CHAPTER 15

TIMBER STRUCTURES
Donald W. Neal, P.E.

ENGINEERING PROPERTIES OF WOOD Three Distinct Grain Orientations Establishing Design Values Bending Tension Parallel to Grain Tension Perpendicular to Grain Compression Parallel to Grain (Short Compression) Buckling (Long Compression) Compression Perpendicular to Grain Shear Stiffness (MOE) CODES, SPECIFICATIONS, STANDARDS, AND REFERENCE PUBLICATIONS Model Building Codes National Design Specification (NDS) AITC Standards AITC Timber Construction Manual (TCM) ANSI/AITC Standard 190.1 and AITC 200-92 Inspection Manual Grading RulesLumber and Timber Wood Handbook: Wood as an Engineering Material Evaluation, Maintenance, and Upgrading of Wood Structures Western Woods Use Book STRUCTURAL TIMBER CHARACTERISTICS AND STRESS ADJUSTMENTS Organic MaterialMoisture Content Variability Time-Dependent Relationship Size/Volume Effect Temperature Adjustment Factors STRUCTURAL MEMBER TYPES Beams Columns Trusses Arches Domes METHODS OF TESTING Wood Moisture Content Coring Stress Wave

15.2 15.3 15.4 15.5 15.5 15.5 15.6 15.6 15.7 15.7 15.7 15.8 15.8 15.8 15.8 15.9 15.9 15.9 15.9 15.10 15.10 15.10 15.10 15.12 15.12 15.13 15.14 15.14 15.16 15.16 15.19 15.19 15.20 15.20 15.21 15.22 15.22 15.22

15.1

15.2

CHAPTER FIFTEEN

Chemical Testing Slope of Grain Bending Tension Compression Shear Delamination Modulus of Elasticity Connector Testing Full-Structure in Situ Load Testing TYPES AND CAUSES OF NONPERFORMANCE AND FAILURE Evolution of Engineered Timber Design War Production Directive Radial Tension Allowable Design Stress Loading/Overloading Dead Load Snow and Snow Drifting Ponding Wind Load Unbalanced Load Deterioration Decay, Termites and Marine Borers Chemical Deterioration Heat Lateral Buckling Notching or Tapering Tension Face Eccentricity Secondary Effects State-of-the-Art Design Changes Compatibility of Analysis and Structural Performance STRUCTURAL EVALUATION OF IN SITU TIMBER STRUCTURES TEMPORARY AND PERMANENT REMEDIAL REPAIRS Shoring Replacement Sistering Bolt Tightening Clamps and Stitch Bolts Beam Stretcher Moment Connections Post Tensioning Radial Tension/Cross-Grain Tension Reinforcement Shear Reinforcement EXAMPLES AND CASE STUDIES REFERENCES

15.22 15.23 15.23 15.23 15.24 15.24 15.24 15.24 15.24 15.24 15.25 15.25 15.25 15.26 15.26 15.27 15.27 15.28 15.28 15.28 15.29 15.29 15.29 15.30 15.30 15.30 15.30 15.31 15.31 15.31 15.32 15.33 15.34 15.34 15.34 15.34 15.35 15.35 15.35 15.35 15.36 15.36 15.37 15.57

ENGINEERING PROPERTIES OF WOOD Wood and brick and stone have been the construction materials of choice since antiquity. Development of wood as an engineered material based upon known

TIMBER STRUCTURES

15.3

design properties, rather than rules of thumb and artisans judgment, is more recent. The 20th century has seen virtually all this development with an accelerating pace toward the end of the century. Wood fibers are formed in the living tree with their long dimension approximately parallel to the axis of the tree trunk. Wood is classified as anisotropic in that it has distinct and different mechanical properties in the three major axes. (See Fig. 15.1.)
Three Distinct Grain Orientations.

Longitudinal Radial Tangential

Parallel to grain Normal to annual rings and perpendicular to grain Tangent to annual rings and perpendicular to grain

Wood grain direction is the wood fiber orientation. For engineering purposes, design properties are referred to as parallel or perpendicular to the grain. This means approximately parallel or perpendicular to the faces of finished lumber or timbers. The wood fiber is not precisely parallel or perpendicular to the faces of finished lumber because of natural growth characteristics or sawmill tolerances. The angle between wood fiber orientation and the faces of finished lumber is termed the slope of grain, and its impact upon strength and stiffness is addressed in the lumber-grading rules. The annual rings are formed radially owing to wood density differences in the growth seasons. In a finished piece of lumber, the orientation of annual ring lines is not necessarily parallel to the wood fiber or grain orientation. Checking that develops as wood dries will be parallel to the grain orientation. Anyone who has split firewood understands that wood is anisotropic, not isotropic, like many other materials. Since the difference between radial and tangential annual ring orientation in a given piece of lumber is normally random and the difference in properties between radial and tangential orientation is relatively

FIGURE 15.1

The three principal axes of wood with respect to grain direction and growth rings. Longitudinal: parallel to grain; radial: normal to annual rings and perpendicular to grain; tangential: tangent to annual rings and perpendicular to grain. (From Wood Handbook, USDA Forest Service.)

15.4

CHAPTER FIFTEEN

small, the engineer needs only to consider properties perpendicular or parallel to the grain for design purposes. Distinctly different properties between perpendicular and parallel to the grain include the modulus of elasticity (MOE), shear, tension, compression, and dimensional change due to wood moisture content variation. Traditionally the basic allowable stress design (ASD) values for lumber and timbers have been determined per the clear-wood procedure. Small, clear, straight-grained specimens of a given species are tested in bending, tension parallel to grain and compression parallel to grain in sufficient quantity for statistical analysis. A series of factors are then applied which include adjustment for size effect, moisture content, safety factor, a strength ratio between the value for clear wood and the grade for which a basic allowable stress is desired, and other adjustment factors where applicable. Basic allowable stress design values for solid sawn timbers are still established using the clear-wood procedure. Following an extensive testing program, allowable design values for dimension lumber (nominal 2- to 4-in thickness) are now established by the in-grade procedure. This program consisted of more than 70,000 pieces of full-sized lumber from production runs tested in bending, tension and compression parallel to grain, and allowable design properties subsequently determined per American Society for Testing and Materials (ASTM) procedures. This large data set was used to develop a model to assign design values for North American dimension lumber. Tabular stresses resulting from the in-grade program reflect actual lumber in the marketplace rather than statistical theory. The program also addressed the current status of the material source from large log to small log mills and old growth to second-growth resource which has changed over time. The 1991 National Design Specification1 (NDS) first presented allowable dimension lumber stresses based upon the in-grade procedure. Over the past several years, industry trade and professional associations have worked to produce a Load Resistance Factor Design (LRFD) Manual of Wood Construction, coordinated by the American Forest and Paper Association/American Wood Council. The manual is essentially an LRFD version of the National Design Specification. Evolution of concrete design from ASD to ultimate strength design occurred over a considerable period, and evolution of steel design from ASD to LRFD is ongoing. Acceptance of LRFD for wood construction by code jurisdictions and the design community is not yet determined. Glulam timber is a sandwich of bonded laminates whose design properties are dependent upon the properties of these component laminates. Since most glulam is bonded from dimension lumber sizes, the basic design values were initially those of its dimension lumber laminates. In the late 1960s, an ongoing testing program was initiated by the American Institute of Timber Construction and the Forest Products Laboratory, using full-sized glulam members. Basic design values in bending and tension were back-calculated from these test data, causing a reduction in allowable stresses. The current design procedure incorporating the backEstablishing Design Values.

TIMBER STRUCTURES

15.5

calculated test results is based upon statistical procedures in accordance with ASTM D 3737.2 A procedure is published by the American Institute of Timber Construction (AITC) in AITC-500,3 which is based upon ASTM D 3737. This procedure permits computation of basic allowable design values for virtually any combination of laminates within a glulam member. While both ASTM D 3737 and AITC-500 documents are available, the computation of allowable design values using this procedure is both time-consuming and tedious. These computations have been computerized by the trade associations servicing the glued laminated timber industry. A majority of timber structural members are used primarily in bending, so bending is often considered the defining design property. Basic bending stress is often used to label the grade for both solid sawn and glulam members. Within the elastic range, the flexural stress block is considered to be a straight line, and extreme fiber bending stresses are calculated as MC /I. The modulus of rupture (MOR), by definition, is the value of MC /I calculated using the bending moment M at ultimate load. The MOR is not the maximum fiber stress at failure since the stress block beyond the elastic range is not a straight line. However, the MOR is a useful index and is often used for test comparisons.
Bending. Tension Parallel to Grain.

Until recent years, test equipment was not readily available to pull full-sized pieces of lumber to failure in tension, so only small striptension tests were used. Older tabulated allowable design values for both solid sawn timbers and glulam grouped bending and tension parallel to grain together on the theory that pure tension stress would behave the same as the tension portion of bending stress. The 1968 NDS was the first to provide separate columns of allowables with different design values for bending and tension.

Tension perpendicular to grain, also called crossgrain tension, describes stress perpendicular to the wood fiber. This may result from an applied gravity load at the lower edge of a member, or from a lateral load such as that applied to a ledger. When tension perpendicular to grain is induced by an increase in radius, straightening the arc of a curved beam, it is commonly termed radial tension. Cross-grain tension is the property of timber least understood among design professionals. Cross-grain tension is not presently tabulated in allowable design tables. Allowable values for radial tension are covered in the body of NDS and in AITC specification 117.4 The basic radial tension allowable stress is 15 lb/in2 for Douglas fir-larch, Douglas fir-south, hem-fir, western woods, and Canadian softwood species subjected to gravity loading. The low value of 15 lb/in2 was assigned on the basis of judgment and field experience. The allowable radial tension stress has historically been one-third of the allowable horizontal shear stress for the above species subjected to lateral loading and for southern pine subjected to all
Tension Perpendicular to Grain.

15.6

CHAPTER FIFTEEN

types of loading. (See Case Study 11, Lucky Strike Lanes, and Case Study 15, Hood River Valley High School Theater.) This is termed precisely crushing strength under the provisions of ASTM Standard D 1985 and is defined as the maximum stress sustained in compression parallel to the grain by a specimen having a ratio of length to least dimension of less than 11. Compression parallel to grain tests are often possible to obtain during forensic evaluation, as a relatively small test specimen is required and the test setup is simple. From this test compressive MOE, proportional limit and ultimate compression data may be obtained. While useful as comparative data in evaluation, allowable stresses in compression parallel to grain are of limited value to the design engineer as most compression member capacity is limited by buckling.
Compression Parallel to Grain (Short Compression). Buckling (Long Compression). Buckling strength of timber members in compression is defined by the Euler critical buckling stress and the slenderness ratio 1/d, where l is the distance between points of lateral support and d is least cross-sectional dimension. From the 1950s until the 1977 NDS, the allowable column stress was the lesser of the crushing strength or the following Euler buckling equation, which is adjusted for factor of safety:

Fc

0.3 E (l/d)2

where E modulus of elasticity (MOE). From the 1977 NDS until the 1991 NDS, columns were classified as short, intermediate, and long. Short columns Long columns Columns with an l/d ratio of 11 or less used an allowable stress based upon crushing strength. Columns with an 1/d ratio greater than a K factor, defined in the specification, used the adjusted Euler buckling stress design equation. Columns with an l/d ratio between 11 and K used a formula representing interaction between crushing and the adjusted Euler buckling stress design equation.

Intermediate columns

The 1991 NDS incorporated the single continuous column formula, often referred to as the Ylinen formula, for all ranges of slenderness ratio with interaction between crushing and Euler buckling failure modes. The continuous column formula is a refinement of the Euler equation with different input constants dependent upon the type of timber member (sawn lumber, glulam, etc.) FC FC* Cp FC * 1 FCE /FC* 2C 1 FCE /FC* 2C
2

FCE /FC* c

TIMBER STRUCTURES

15.7

where FC* FCE

KCE

c le

tabulated compression design value multiplied by all applicable adjustment factors except Cp critical buckling design value for compression members KCE E (le/d)2 Euler buckling coefficient for columns 0.3 for sawn lumber 0.418 for glulam 0.8 for sawn lumber 0.9 for glulam effective column length

Compression perpendicular to the grain is used primarily to size the required bearing area of beams over supports or under concentrated loads. Failure in compression perpendicular to the grain cannot be as well defined as with other modes of failure. Excessive compression perpendicular to grain loading causes wood cells to collapse and crush but rarely causes member collapse or catastrophic failure. Prior to the 1982 NDS, allowable stresses were based upon a proportional limit approach. Starting with the 1982 NDS, the basis for the allowable design value in compression perpendicular to the grain was revised and now is tabulated as the mean stress at a stated deformation. Deformation limits of 0.04 and 0.02 in are commonly used. The time-dependent load duration factor CD is not applicable to a property defined by the deformation limit and is not used with compression perpendicular to grain calculations.
Compression Perpendicular to Grain.

Tabulated allowable shear stresses are for the shear plane parallel to the faces of finished lumber and approximately parallel to wood fiber orientation. Since most timber members are beams oriented horizontally, this is commonly termed horizontal shear. For the most part, allowable shear stress is unchanged between grades of a given specie. Data obtained by shear testing of glulam per ASTM D 198 resulted in increased allowable shear stresses being adopted in 1994 and additional increases in 1998. The 1994 revisions are incorporated in the 1997 NDS, and both 1994 and 1998 revisions are available as addenda to AITC Specification 117 Design. It may be some time before all these revisions find their way to the structural design community by way of incorporation into the model building codes.
Shear. Stiffness (MOE). Youngs modulus, or the modulus of elasticity (MOE), is the slope of the stress-strain curve and may be measured in the testing laboratory when members are loaded in flexure or axial tension or compression. The tabulated MOE included in tables of allowable design stresses is an average value. MOE in bending is included with all design property tables for dimension lumber, timbers, and glulam. Glulam design tables include MOE in bending for both axes of the laminations, and MOE in axial loading. Machine stress-rated (MSR) grades of dimension lumber have

15.8

CHAPTER FIFTEEN

MOE measured for each piece at the point of manufacture. The MOE rating and an accompanying allowable bending stress are used for member identification, and tabular values are listed in table 4C of the NDS Supplement.

CODES, SPECIFICATIONS, STANDARDS, AND REFERENCE PUBLICATIONS The three model U.S. building codesUBC, BOCA, and Standardare scheduled for merger in the near future. Each city, county, or state jurisdiction selects and adopts a given code as of a given date. The code typically incorporates specifications already developed, such as the NDS, but differences may exist on key points. For example, the 1994 UBC adopted the 1991 NDS except for a difference in the load duration factor CD for lateral wind loading. It is common for a given jurisdiction to add or revise portions of the model code, tailoring it to specific climatic or site conditions such as permafrost, hurricanes, expansive soils, and snow. It is imperative in forensic evaluation to know the specific code and code revisions that a structure must be evaluated against. The building code is the legal document for structural compliance, and if differences exist between the code and other specifications or standards, the code prevails. The code is a minimum standard, and the prudent engineer may incorporate more conservative criteria of which he/she has knowledge, if those criteria have not yet been adopted into the code.
Model Building Codes.

The National Design Specification for Wood Construction may be obtained from the American Forest & Paper Association at 1111 19th Street N.W., 7th floor, Washington DC 20036. This is the source document for design with solid sawn timber and its connections. It was first published during World War II and was periodically updated with editions in 1944, 1957, 1960, 1962, 1968, 1971, 1973, 1977, 1982, 1986, 1991, and 1997. The main volume contains design provisions and equations for design of solid sawn and glulam members, plus provisions and design allowables for connections. The Supplement contains section properties and allowable design values for both solid sawn and glulam members. The Commentary provides a historical overview and explanation behind the specification provisions.
National Design Specification (NDS).

American Institute of Timber Construction (AITC) [at 7012 S. Revere Parkway #140, Englewood, CO 80112] publishes the following standards, designated by AITC number and title:
AITC Standards.

104 108 109

Typical Construction Details Standard for Heavy Timber Construction Standard for Preservative Treatment of Structural Glued Laminated Timber

TIMBER STRUCTURES

15.9

110 111 112 113 114 117 119

Standard Appearance Grades for Structural Glued Laminated Timber Recommended Practice for Protection of Structural Glued Laminated Timber During Transit, Storage and Erection Standard for Tongue-and-Groove Heavy Timber Roof Decking Standard for Dimensions of Glued Laminated Structural Members Structural Glued Laminated Timbers for Electric Utility Framing and Crossarms Standard Specifications for Structural Glued Laminated Timber of Softwood Species Standard Specifications for Hardwood Glued Laminated Timber

All the standards are potentially valuable to the forensic engineer. The most commonly used of the standards, and the source document for design with softwood glulam timber, is Specification 117, Design, which provides allowable design values and design examples. The latest edition was in 1993 with addenda in 1994 and 1998. The 117 specification is issued in two documents, Design and Manufacturing. The manufacturing document showing lamstock grade requirements and layup zones for the various glulam combinations is of interest primarily to those in glulam manufacture.
AITC Timber Construction Manual (TCM). The AITC Timber Construction Manual6 is published by John Wiley & Sons, Inc. At the time of this writing the latest volume is the 1994 fourth edition, which was preceded by editions in 1966, 1974, and 1985.

ANSI/AITC Standard 190.17 for wood products, Structural Glued Laminated Timber, is a consensus standard concerned primarily with the nuts and bolts of glulam manufacture. The standard is revised periodically, the current edition being 1992. The AITC 200, Inspection Manual,8 available from the same source, describes all the tests and procedures for qualification, production, and testing of glulam timber, and is based upon ANSI/AITC 190.1.
ANSI/AITC Standard 190.1 and AITC 200-92 Inspection Manual.

Rules-writing agencies have evolved which develop lumber grades for specific species within a geographic area. Lumber mills producing graded lumber join an appropriate grading agency, and their products carry that agencys stamp.
Grading RulesLumber and Timber.

The Wood Handbook9 is available from Forest Products Society at 2801 Marshall Court, Madison, WI 53705. The handbook contains a wealth of technical data on wood not found elsewhere. The 1999 edition was preceded by editions in 1935, 1955, 1974, and 1987, all by
Wood Handbook: Wood as an Engineering Material.

15.10

CHAPTER FIFTEEN

the U.S. Government Printing Office. The 1999 edition is published by the Forest Products Society.
Evaluation, Maintenance, and Upgrading of Wood Structures.

The American

Society of Civil Engineers (ASCE) published this in 1982.

Western Woods Use Book.

This is published by Western Wood Products Association (WWPA) in 1996.

STRUCTURAL TIMBER CHARACTERISTICS AND STRESS ADJUSTMENTS The basic design strengths in structural timber are influenced by numerous factors which include natural growth characteristics, wood moisture content, temperature, duration of design load, size/volume effect, loading orientation, and lateral stability. Adjustment factors have been developed to adjust the basic tabular design stresses for these conditions when applicable. (See Table 15.1.) Wood is a natural-growth organic material with a cellular structure, which retains water. Water is a living trees best friend and lumber products worst enemy. Wood moisture content is expressed as a
Organic MaterialMoisture Content. TABLE 15.1 Applicability of Adjustment Factors Load duration factor Fb Ft Fv Fc Fc E Fg
a b

Wetservice factor CM CM CM CM CM CM

Temperature factor Ct Ct Ct Ct Ct Ct Ct

Beam stability factor a CL

Size factor b CF CF CF

FlatVolume use factor a, c factord CV Cfu

Fb Ft Fv Fc Fc E Fg

CD CD CD CD CD

The beam stability factor CL shall not apply simultaneously with the volume factor Cv for glued laminated timber bending members. Therefore the lesser of these adjustment factors shall apply. The size factor CF shall apply only to visually graded sawn lumber members and to round timber bending members. c The volume factor Cv shall apply only to glued laminated timber bending members. d The flat-use factor Cfu shall apply only to dimension lumber bending members 2 to 4 in (nominal) thick and to glued laminated timber bending members. e The repetitive-member factor Cr shall apply only to dimension lumber bending members 2 to 4 in thick.

TIMBER STRUCTURES

15.11

weight percentage of the water present to the dry weight of wood. Water is retained in the cell walls to saturation, and additional free water may be retained within the cell cavity. The fiber saturation point of most wood species occurs in the middle 20 percent range with Douglas fir about average at 25 to 26 percent. Additional moisture above the fiber saturation point may be retained as free water within the cells. Since water softens the cellular wall, strength and stiffness increase as moisture content is reduced below the fiber saturation point. Design values for solid sawn softwood lumber are based upon wood moisture content of 15 percent described as an average of 19 percent or below, with a wet-service factor CM reduction applied where moisture content will exceed 19 percent. Glulam is usually produced in a moisture content range of 12 to 15 percent. Design values for glulam are based upon wood moisture content of 12 percent with a wet-service factor CM reduction applied where moisture content will exceed 16 percent. Wood in use and not subject to free water will come to equilibrium at a moisture content as a function of temperature and humidity. An oven-dry laboratory test in which all moisture is driven off can accurately determine the weight of water in a specimen, from which the wood moisture content may then be calculated. A simpler but not quite so precise method employs an electric resistance wood moisture meter. A moisture meter and probe, such as the G-30 by Delmhorst Instrument Co., should be in the forensic engineers tool kit. Adjustments for specie and temperature are provided by the instrument supplier. Wood moisture content in the higher ranges provides an environment favorable to wood-eating

Incising factor Ci Ci Ci Ci Ci Ci

Repetitivemember Curvature factorf factore Cr Cc

Form factor Cf

Column stability factor CP

Shear stress factorg CH

Buckling stiffness factorh CT

Bearing area factor Cb

The curvature factor Cc shall apply only to curved portions of glued laminated timber bending members. Shear design values parallel to grain Fv for sawn lumber members shall be permitted to be multiplied by the shear stress factors CH. h The buckling stiffness factor CT shall apply only to 2-in 4-in or smaller sawn lumber truss compression chords subjected to combined flexure and axial compression when 3 8-in or thicker plywood sheathing is nailed to the narrow face. Source: American Forest & Paper Association, Washington, D.C.
g

15.12

CHAPTER FIFTEEN

organisms, fungi, termites, etc. A softwood moisture content of 20 percent is considered the approximate threshold at which primitive forms of life that cause decay become active. Many of the problems a forensic engineer is asked to investigate with wood structures are moisture-related. All structural materials have characteristic variability from one member to another. Wood has natural growth characteristics that contribute to its variability. Allowable design stresses are based upon adjustments for safety factor and load duration that are applied to 5th percentile strength values, thus accounting for the variability. Glulam may have an allowable bending stress nearly double that of solid sawn lumber of the same species. By dispersion of defects and reforming the wood using graded laminates, the variability of the end product is reduced. Structural members of smaller glued veneer-size laminates, such as laminated veneer lumber (LVL) and parallel-strand lumber (PSL), further reduce variability. The coefficient of variation (COV) for MOE of wood products is approximately as listed:
Variability.

0.40 0.25 0.11 0.10 0.06

ungraded solid sawn lumber visually graded solid sawn lumber machine stress-rated solid sawn lumber Glulam LVL, PSL, etc.

A frequency histogram for MOE would consist of a bar chart with the Y axis height of bars indicating the frequency of occurrence of each increment of MOE which is plotted on the X axis. The histograms shown in Fig. 15.2 for the five COVs listed above are a smoothed-out version of the bar charts.
Time-Dependent Relationship. A unique property of structural wood is its ability to sustain higher loads for a short time than under extended or permanent loading. Tabulated allowable design stresses are for normal loading duration defined as a 10-year period. Design stress adjustments must be applied for other loading durations, as shown in Table 15.2. Be aware that building codes for some jurisdictions may require a different load duration factor CD for wind loading than that listed in the NDS and that the building code always governs. The load duration is the accumulated time the design load is applied during the life of the structure. For example, the usual duration factor for snow loading is 1.15 for a 2-month duration. This does not mean 2 months of snow loading per year or 2 months of continuous snow loading, but rather 2 months of design snow loading during the life of the structure. If the design snow load could reasonably be applied for more than 2 months, such as at a high elevation or a ski resort, then the 1.15 factor is not appropriate and should be reduced.

TIMBER STRUCTURES

15.13

1.6 1.4 0.25 0 0.5 1 1.5 2 Ungraded MOE (a) 2.5 3 0 0.5 0.94 1 1.5 2 VSR MOE (b) 2.5 3

COV = 0.50

E.05/E = 0.18

COV = 0.25

E.05/E = 0.59

1.7

1.8

1.39 0 0.5 1 1.5 MSR MOE (c) 2 2.5 3 0 0.5 1

1.5 1.5 2 Glulam MOE (d) 2.5 3

COV = 0.11

E.05/E = 0.82

COV = 0.10

E.05/E = 0.84

2.0

1.8 0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 All MOE Distributions (f) 3

COV = 0.06

PSL/LVL MOE (e)

E.05/E = 0.90

FIGURE 15.2

Frequency histograms for various structural wood products.

Size/Volume Effect. Testing over a long time confirms that smaller wood members have a higher modulus of rupture (MOR) than larger members. This appears to be true across the full range of member sizes. Research at the Forest Products Laboratory resulted in a depth factor formula as a function of depth only in use until the middle 1960s. In 1966 a size factor CF as a function of depth only was adopted as a modifier to allowable bending stress for both solid sawn lumber and timbers plus glulam:

15.14

CHAPTER FIFTEEN

TABLE 15.2 Frequently Used Load Duration Factors Load duration Permanent 10 years 2 months 7 days 1 day 10 min Instantaneous CD 0.9 1.0 1.15 1.25 1.33 1.6 2.0 Typical design loads Dead load Occupancy live load Snow load Roof live load Wind/seismic load Wind load Impact load
1

CF

12 d

The size factor remains in effect for solid sawn timbers as a modifier to bending, tension parallel and compression parallel tabular stresses and is also used as a modifier to bending for LVL and PSL products. In 1991 the volume factor Cv as a modifier to bending stresses was adopted for glulam which is a function of depth, width, and length. The basic formula shown below is further modified for support and loading conditions by a KL factor: CV where x L d b KL KL 21 L
1/x

12 d

1/x

5.125 b

20 for southern pine 10 for other species beam length, ft beam depth, in beam width, in loading condition coefficient furnished in NDS

The volume factor is more conservative for larger glulam members than the previous size factor.
Temperature. The tabular design stresses for wood are based upon a temperature of approximately 20C with higher strengths at low temperature and lower strengths at high temperature. No allowable stress increase is permitted for low-temperature use, but Table 15.3 from NDS provides reduction factors for wood used at elevated temperatures. Up to approximately 150F for short exposure times the temperature effect is reversible. At temperatures above 150F or at temperatures above 100F for repeated or prolonged periods, some of the strength loss is permanent. Adjustment Factors. A common error in design or evaluation of structural timber is the use of tabulated design values without adjustment of those values for specific

TIMBER STRUCTURES

15.15

TABLE 15.3 Temperature Factor Ct Ct Design values Et , E Fb , Fv, Fc , and Fc In-service moisture conditions Wet or dry Dry Wet

100F 1.0 1.0 1.0

100F

T 0.9 0.8 0.7

125F

125F

T 0.9 0.7 0.5

150F

Source: American Forest & Paper Association, Washington, D.C.

use conditions. The list of adjustment factors for design values has grown over time, and use of the adjustment factors is often perceived as a confusing complication in timber design. All have a rational basis in design refinement. In practice, most designs use only one to three adjustment factors. Several factors are used only in specialized applications. The Applicability of Adjustment Factors table from the NDS, furnished here as Table 15.1, provides a quick review of the adjustment factors required for a specific design. A similar table is provided in NDS for applicability of adjustment factors required for connection design in timber. Load Duration Factor CD. This factor addresses the time-dependent property of timber to sustain a given loading. (See Table 15.2.) Factor CD is applied to all tabulated design values except MOE, which is not time-dependent, and compression perpendicular to grain which is based upon a deformation limit. The load duration factor is used in virtually all design and evaluation calculations. (See the paragraph Time-Dependent Relationship.) Wet-Service Factor CM. (See paragraph entitled Organic MaterialMoisture Content.) The wet-service factors are furnished at the front of design tables in the NDS for solid sawn lumber and timbers. The wet-service factors are furnished at the lower edge of tabular design tables for glulam in AITC Specification 117. Temperature Factor Ct. The temperature factor is rarely used in design as it applies only to members subjected to temperatures above 100F for repeated and prolonged periods. It should be considered in forensic evaluation of members subjected to elevated heat such as plywood dryers, papermills, industrial or food processing applications, and attics in hot climates. (See paragraph entitled Temperature and Table 15.3.) Beam Stability Factor CL. The beam stability factor was introduced with the 1991 NDS and is applied only to allowable bending stress. It addresses the buckling tendency of beams where the compression edge is not laterally stabilized. In the design of solid sawn lumber, CL is applied simultaneously with the size factor CF. In the design of glulam, CL, which is a function of the compression edge, is not applied simultaneously with volume factor CV, which is a function of the tension edge. The lesser of the two factors applies.

15.16

CHAPTER FIFTEEN

Size Factor CF. The size factor applies to bending, tension and compression parallel to grain, for solid sawn lumber and timbers. The factors are furnished at the front of design tables in the NDS. (See paragraph entitled Size/Volume Effect.) Volume Factor CV. The volume factor applies only to bending stress for glulam. The volume factor is not applied simultaneously with the beam stability factor CL. The lesser of the two factors applies. (See paragraph entitled Size/Volume Effect.) Flat-Use Factor Cfu. The flat-use factor applies to bending stress for members with load applied to the wide face of member. The factors for dimension lumber are furnished at the front of design tables in the NDS. In glulam members a flatuse factor is applied when loading is parallel to the wide face of the lamination. The factors are furnished in AITC Specification 117. Repetitive-Member Factor Cr . The repetitive-member factor applies only to bending stress for dimension lumber to account for load sharing by sheathing elements. The 1.15 factor may be applied to a system of at least three members spaced at not more than 24-in centers. Curvature Factor Cc. The curvature factor applies only to bending stress at curved glulam members per the formula

Cc where t
Form Factor Cf .

1 2000 (t/R)2

lamination thickness and R radius of curvature. This rarely used factor is applied only to bending stress. It converts round or diamond-shaped cross-section bending members to the same moment capacity as a square member of the same cross-sectional area. Column Stability Factor Cp. Similar to the beam stability factor and introduced with the 1991 NDS, the factor addresses buckling tendency and is applied only to compression parallel to grain design stresses. Shear Stress Factor CH. The allowable shear stress of solid sawn lumber is lower than glulam to account for separations such as splits, checks, and shake. Factors at the front of NDS design tables for dimension lumber and timbers permit increases in allowable shear stress dependent upon the measured size of separation in a specific member. This factor should not be used for design, but is of value to the forensic engineer for evaluation of specific members. Incising Factor Ci . When sawn lumber is incised with incisions parallel to grain to increase treatment penetration, an adjustment to MOE and allowable design values is required.

STRUCTURAL MEMBER TYPES A majority of structural lumber and timber products find use as bending members such as girders, stringers, beams, purlins, joists, rafters, and decking. Virtually all solid sawn and most glulam bending members are prismatic, having a constant cross section. The development of glulam enlarged the possibilities of
Beams.

TIMBER STRUCTURES

15.17

deeper, wider, and longer sections in straight, cambered, or curved shapes. See Fig. 15.3 for timber beam profiles. Tapering the cross section to a nonprismatic shape is done to form the straight tapered or the popular curved pitched and tapered beam profiles. Since wood is composed of longitudinal fibers, tapering of the cross section severs stressed wood fibers and causes a readjustment of stresses carried by those fibers. The AITC Timber Construction Manual (TCM) contains a design procedure for both straight tapered and curved pitched and tapered beams which is substantially more complex than the design of prismatic beams. Severing of wood fibers stressed in tension is particularly harmful structurally, so any tapering should be confined to the compressive edge of members. In a glulam member, the laminations should be parallel to the nontapered edge. Since glulam is a structural sandwich normally laid up with higher grades of material near its top and bottom flanges, a reduction in allowable stress may be required if the taper cut removes the higher-grade zone, exposing the lower-grade core material to maximum bending stress. Custom layup combinations can provide higher-grade laminates throughout the tapered portion.

STRAIGHT PRISMATIC

STRAIGHT TAPERED

d CURVED PRISMATIC

dn

NOTCHED

CURVED PITCHED AND TAPERED

B L

D L

L 3-SPAN CANTILEVERED

FIGURE 15.3

Timber beam profiles.

15.18

CHAPTER FIFTEEN

Notching of timber structural members also severs stressed wood fibers. A combination of section loss, stress concentrations at the notch, and interaction of tension perpendicular to grain and shear stresses create a complex situation for analysis. Knowledge of fracture mechanics and stress concentrations beyond that possessed by most practicing structural engineers is required to understand the complex situation created by notching timber flexural members. The following formula for shear stress at a tension face notch has been used since early versions of the NDS: fv where v b d dn shear at notch location member width full member depth member depth at notch 3v 2bdn d dn

The formula is based upon testing of short, relatively deep beams with a span-todepth ratio of 9. Caution should be used in reliance on the above formula, which does not address characteristics and sharpness of the notch, the possibility of separations or low strength perpendicular to grain in the shear plane, or beam profiles with a span-to-depth ratio greater than 9. Over the years due to service problems with notched beams, more restrictions have been added. The NDS limits notch depth at the tension face for solid sawn members to d/4 at a support and to d/6 elsewhere. The AITC Timber Construction Manual limits notch depth at the tension face to d/10 for glulam beams. The Uniform Building Code prohibits sawn taper cuts on the tension face of glulam beams. In-service problems suggest that beams tapered or notched at the tension face should be avoided. (See Case Study 4, First Nazarene Church.) Cantilever beam systems are commonly used for roof framing of warehousetype structures. Negative bending moment, caused by downward force of the link span at the end of a cantilever beam, permits a smaller more economical beam section (see Fig. 15.3). Published tables and formulas usually give optimum cantilever length, assuming balanced loading. Unbalanced span loading should be checked in design or evaluation of cantilever beam systems. (See Case Study 1, Port of Longview Warehouse No. 1.) The optimum cantilever length is that at which bending moment at support C for the balanced-load condition is equal to maximum moment near midspan B for the unbalanced-load condition (see Fig. 15.3). Experience with cantilever beam systems suggests that the economic benefits often vanish when analyzed for unbalanced span loading. Deflection should be checked as the reduced section, based upon bending criteria alone, may result in a beam too flexible and subject to ponding. (See Case Study 1, Port of Longview Warehouse No. 1.) The portion of beam past the inflection point where compression is at the lower portion of beam should be considered in regard to buckling stability. One problem with cantilevered glulam beams is that the deflected shape is complex, and

TIMBER STRUCTURES

15.19

it may be preferable to use a straight beam, rather than attempting to camber. A recent development in glulam technology utilizes high-strength composite fibers bonded into the tension zone of a glulam beam. The fiber panels are fiberglass, kevlar, or similar-type material. The composite panel is bonded to the beam tension face with a wood coverboard for cosmetic purposes. Strength of the tension zone is increased with the goal of reducing member size. Addition of the composite reinforcing may alter the member sizing from strength-controlling to stiffness-controlling. Availability of composite-reinforced glulam at this point is through proprietary suppliers.
Columns. Timber is a good material for columns in that allowable stresses are high in compression and the relatively larger cross section compared to steel limits the buckling tendency. Column design is based upon the continuous column formula, which is an interaction between crushing strength of pure compression and Euler buckling.

Wood trusses may be roughly divided into heavy truss and light truss categories. Heavy trusses are typically spaced from 12 to 20 ft apart, have intermediate joist or purlin framing between them, and are designed by a structural engineer and custom-built. Light trusses are typically spaced from 12 to 48 in apart, are covered with sheathing, and may have the benefit of load sharing. They are typically designed and manufactured as proprietary products. Failure at any chord, web member, or joint in a truss can cause failure of the full truss, so a more detailed analysis is required than for a beam. Older truss designs, which the forensic engineer is often called upon to evaluate, usually ignored joint and heel eccentricity and chord continuity which could not be addressed by approximate precomputer methods of analysis. Bowstring trusses are among the most efficient truss forms in timber. Almost a variation of a curved tied arch, the web forces are very small for balanced loading. Top chord bending is also relatively small because the positive moment from gravity loading is partially offset by the negative Pe moment, which is the product of top chord compression force and the eccentricity due to curvature. Web-to-chord connection forces are relatively small. Most of the in-service problems with bowstring trusses involve lack of safety factor for overloading or state-of-the-art reductions in allowable tension stresses at the lower chord. (See Case Study 9, Hughes Aircraft Building, and Case Study 10, Sheridan High School Gym.) By contrast, heavy parallel chord trusses have large web forces, large web members, and large web-to-chord connections. Most trusses are unstable alone and require bracing, both during construction and in service, to remain structurally viable. In most cases the roof diaphragm of the building stabilizes the truss top chord against lateral buckling. Removal of such bracing elements without replacement bracing can create serious problems. A case in point is a bowstring truss framed industrial building investigated by the
Trusses.

15.20

CHAPTER FIFTEEN

author where roof framing at one-half of the span had been removed and raised to provide natural lighting via a clerestory. It lasted several years until compression caused by a heavy snow loading buckled the unsupported portion of the truss top chords, collapsing the entire building. One advantage of glulam is that it may be produced in curved shapes at a reasonable and competitive cost. A primary application of curved glulam is arches of which the three-hinged Tudor profile is most common (see Fig. 15.4). Although often tapered throughout its full length, the arch is in combined bending and compression so the maximum bending stress is at the compression side of the stress block. As such, it does not usually suffer from the problems of tapering the tension face discussed earlier in this chapter. A graphical equilibrium polygon analysis, considered archaic today for most analysis, provides a rapid procedure for an approximate arch analysis. A precise analysis is complicated by the tapered profile. A full design procedure is provided in the AITC Timber Construction Manual.
Arches.

Glulam timber has distinct advantages as a framing material for spherical domes. Most steel-framed domes are formed with straight members creating a series of flat facets on the dome surface. Glulam dome members may be produced to the great-circle radius of the dome, producing a smooth, spherical surface which is less costly to sheathe and roof. Domes of the reticulated-type pattern are a series of triangles on the domical surfacea three-dimensional truss. Curved glulam members carry primarily compression with bending relatively small, as the positive moment from gravity loading is partially offset by the negative Pe moment. Most dome analysis consists of a three-dimensional space truss approach where roof sheathing is considered only to provide lateral buckling restraint to the primary members. Shell action of the sheathing is usually ignored and provides some additional safety factor. Connections are designed per the analysis, with joint tension caused by concentrated or unbalanced loading. Any pattern of space truss members on the dome surface, which may be defined, analyzed, designed, and connected, may be used to frame a dome. The designer strives for a pattern pleasing to the eye, which is efficient and
Domes.

FIGURE 15.4

Tudor arch profile.

TIMBER STRUCTURES

15.21

LINE ON GREAT CIRCLE OF SPHERE PATTERN PLANE

RISE

PATTERN CIRCLE SPHERICAL DOME SHAPE

DIAME

TER

RADIUS CENTER

DOME SHAPE SPHERICAL SECTION

ROOFING W.P. RISE SHEATHING W.P.

DIAMETER

DOME FRAMING

FIGURE 15.5

Reticulated domes.

repetitive to design. In a common pattern, a series of equilateral triangles in the plane of supports is projected from the radius center to the dome surface. See Fig. 15.5.

METHODS OF TESTING Standard tests for strength and stiffness in bending, tension parallel to grain, compression parallel to grain, shear and compression perpendicular to grain are specified by ASTM. ASTM D 14310 applies to small specimens, and ASTM D 198 and D 476111 apply to large members. Use of standard tests is strongly recommended.

RADIU

15.22

CHAPTER FIFTEEN

Wood moisture content is expressed as a percentage of the wood moisture weight to the dry weight of wood. A precise measurement of wood moisture content is a destructive test in which a specimen is weighed before and after all moisture is driven off in an oven per ASTM D 4442.12 A nondestructive approximation of wood moisture content may be obtained easily by using a calibrated electric resistance moisture meter where the electric resistance between two needles driven into the wood is measured parallel to grain per ASTM D 4444.13
Wood Moisture Content. Coring.

AITC shear test T-107, per Inspection Manual AITC 200-92, is used to evaluate face bonding adhesive quality and allows two specimen options, one of which is a 1-in-diameter core. To be of value for testing adhesive bond, the cylindrical core must be taken perpendicular to the glue lines. This test is extensively used for in-plant quality control, but is also useful for existing structures when the core may be extracted at a noncritical location. The cores may also be used to calculate material density if it is measured accurately with dial calipers and weighed on a gram scale. Coring or boring with a 3 8- or 1 4-inch bit is a crude, but effective way of quickly determining if advanced decay or deterioration, is present. Lack of drill resistance indicates advanced wood deterioration, and inspection of the shavings indicates if the wood is sound. Powdery or discolored shavings indicate decay or deterioration. Preservative-treated tapered wood plugs are available to seal the coring hole.

Modulus of elasticity is the slope of the stress-strain curve in a static load test. A dynamic MOE may also be determined by the formula
Stress Wave.

MOE where D V g

DV 2 g

material density wave velocity through material constant of gravity, 32.2 ft/s2

Instruments are available with sensors and timers which can measure the elapsed time of a stress wave induced into the material in either a test laboratory or an inplace structure. The wave may be induced with a hammer-type object. If both wave velocity and density can be measured, then dynamic MOE may be calculated. A calibrating factor with static MOE measured by physical testing and dynamic MOE measured by stress wave testing of the same specimen is necessary in order to estimate static MOE using the stress wave procedure. When used to estimate MOE for comparison with design tables, the stress wave is induced parallel to grain. A valuable diagnostic tool for detecting decayed wood is stress wave testing with the wave induced perpendicular to grain. Decayed wood slows the wave velocity, and the location of deteriorated wood may be mapped, using a series of stress wave tests in a gridwork pattern. (See Case Study 7, Hayward Field Grandstand.) Wood is more resistant to chemical action than most construction materials but is not free from chemical deterioration. Acids, acid salts, and
Chemical Testing.

TIMBER STRUCTURES

15.23

alkalines may cause permanent strength loss in wood. Fire-retardant treatment (FRT) chemicals are commonly borates, ammonium sulfate, and ammonium phosphate, which cause wood degradation by acid hydrolysis. During an evaluation, questions often arise regarding the presence of preservative treatment, wood specie, etc. Testing for presence of chemicals is nondestructive as very small test specimens are required. Testing for the presence of preservative treatment is a common and inexpensive procedure. Wood specie identification is available from laboratories or consultants in wood technology. A spectrometer analysis may be utilized to determine the elemental breakdown of retained chemicals. Penetration depth of the chemical in wood is important, and a chemical indicator applied to the specimen is commonly used for that purpose. Some chemicals, such as those used for fire-retardant purposes, are more detrimental near the wood surface due to their interaction with atmospheric moisture. The angle of wood fiber to the longitudinal axis of member, known as the slope of grain, ranks with knots as a natural-growth strength reducer considered when lumber or timber is graded. Electronic instruments are available which can detect slope of grain. A simpler device is a stylus made for this purpose with a sharp needle which follows the wood fiber when pulled by hand in a general parallel to grain direction with the stylus embedded slightly in the wood. The slope of wood fiber should not be confused with the annual ring orientation. The Wood Handbook, in chapter 4, Mechanical Properties of Wood, gives thorough coverage of this topic. (See the paragraph entitled Three Distinct Grain Orientations.)
Slope of Grain.

A static bending test in a laboratory provides a measure of member strength more easily visualized by a client than a ream of calculations. Properly instrumented, it can provide both modulus of elasticity and modulus of rupture. Occasionally a full-sized member is available as a test specimen where one member is sacrificed; or one end of a beam may be available following damage to the other end. More commonly secondary members or strips cut from the member are available as test specimens. Commonly the test setup consists of concentrated loading of the specimen at one-third points, creating uniform bending across the middle one-third of the test beam. AITC test T114, per Inspection Manual AITC 200-92, is a bending test extensively used in the past for testing end joints in individual laminates used for glulam manufacture. Test equipment is now more commonly available for testing end joints in tension.
Bending.

In the past, strips cut from a specimen were usually tested in tension. Equipment is now more commonly available which is capable of pulling specimens in the 2 6 size range to failure in tension. AITC test T119 per Inspection Manual 200-92 is a procedure for pulling a full-size lamination in tension for end joint testing. Where end joint strength is questioned in a glulam forensic inspection and a specimen is available for testing, end joints may be sawn out of the glulam specimen and pull-tested per this procedure.
Tension.

15.24

CHAPTER FIFTEEN

Short column compression crushing strength testing is available from virtually any test laboratory and requires only a small specimen for testing. While compression members rarely function structurally as short columns, this test is inexpensive and can provide compressive MOE in addition to ultimate compressive strength data.
Compression.

AITC test T107, per Inspection Manual 200-92, for glulam members permits the option of stairstep type of block shear tests or core shear tests as specimen options for evaluating shear strength along the face bonding adhesive line. Test machines using a calibrated hydraulic cylinder shear the specimen along the adhesive line. The adhesive bond is evaluated based on adhesive shear strength and an estimate of wood failure percentage in the shear plane. Accurate calibration of the complete system of cylinder and gauge to be used is imperative for meaningful results. Do not accept secondhand information that the test machine is calibrated. Recalibrate rented or borrowed test equipment prior to use. Shear testing presented in a forensic report should be confined to ASTM or industry standard tests.
Shear. Delamination. AITC test T110, per Inspection Manual 200-92, is a cyclic delamination test used for glulam in-plant quality control but is useful for forensic evaluation when possible delamination is under investigation. The test involves several cycles of impregnating a glulam specimen with water in a pressure vessel and rapidly drying in an oven. This test simulates repeated wetting-drying cycles and tests glue line adequacy by placing stress on the glue line. Modulus of Elasticity. Machine stress-rated (MSR) lumber requires measuring the MOE of each piece at the point of manufacture and assigning design properties based upon MOE with visual grading restrictions using high-speed production equipment. In test specimens MOE may be measured by plotting the stress-strain curve from bending, tension, or compression load tests or by stress wave testing. Connector Testing.

The presence of corroded bolts or other fasteners may become the focus of a forensic evaluation, and testing the connector may be required. It may be necessary to determine if a bolt is mild steel or one of the hardened-steel grades. A bolt test for tensile and yield is available at reasonable cost from most test laboratories. ASTM or industry standard tests are recommended. Full-structure load testing is often proposed following evaluation and/or repair of a distressed structure. It is easily visualized, and the structure either passes or fails. Caution should be employed in use of fullstructure load testing for the following reasons: 1. Costs may be substantial. 2. How much load should be applied and for what duration? The load duration factor CD is based upon the assumption that the reserve capacity of the member can withstand a specified load in excess of the design allowable for a specified

Full-Structure in Situ Load Testing.

TIMBER STRUCTURES

15.25

duration during the life of the structure. It is often difficult to determine what percentage of overload to use in a performance test without risking yield or damage to structural elements. 3. It may be difficult to isolate the structure being tested from nonstructural elements fastened to the structure. May elastic deflection be isolated from nonelastic deflection? What if the structural frame tests are satisfactory, but nonstructural walls, gypboard, windows, etc., are damagedwho pays for the damage? 4. When testing is done, be cautious about agreeing to an unrealistic criterion of acceptance. Early in the authors career his company designed and constructed a timber domed structure for a university. The specification required a load test representing full dead plus unbalanced live load, and the criterion of acceptance was that deflection under full load be measured and 80 percent deflection recovery be achieved upon removal of the test load. The structure performed very well under test load. Deflection under full-design loading was substantially less than calculated, but 80 percent recovery was not achieved, as some of the initial deflection was nonelastic due to small amounts of crushing as the connectors seated. As such, the structure did not pass the acceptance criterion, and some negotiation was required prior to acceptance.

TYPES AND CAUSES OF NONPERFORMANCE AND FAILURE

Evolution of Engineered Timber Design

State-of-the-art changes resulting from research and testing caused revisions in the procedure for establishing tabular design stresses, in the magnitude of the tabular stresses, in stress adjustments, and in the analysis procedures. Virtually all these revisions tended to reduce allowable design stresses. Based upon present knowledge, the older tabular stresses for tension and the tension portion of flexural members were nonconservative for an appropriate safety factor. War Production Directive. In August 1943, the U.S. War Production Board issued Directive No. 2914 as an emergency specification, which was simply a revised NDS specifying higher working stresses for lumber, timbers and timber connectors, and defined rules for their use. Issued to help in the war effort, some of the measures were arbitrary and most lacked adequate test data backup. One justification for the measures was short-term use of 5 to 10 years, but many of these World War II structures remain in use today. Lumber and timber working stresses and connector values were increased from previously used values by approximately 20 percent with MOE unchanged. Designs were often based on a roof snow load of 15 lb/ft2 regardless of location. Many of the engineered structures were designed using an allowable bending stress Fb 1200 lb/in2, but the lumber volume required was so large that non-stress-graded lumber was used

15.26

CHAPTER FIFTEEN

without design revision. Structures designed with Fb 1800 lb/in2 lumber were often built with Fb 1200 lb/in2 material. Construction with timber was extremely heavy during this window of time. The wartime directive increase, coupled with the fact that most allowable stresses in timber were higher than current use for tension and bending, is a historical situation the forensic engineer should be aware of when evaluating World War II structures. (See Case Study 9, Hughes Aircraft Building.) Radial Tension. Tension perpendicular to grain is induced in a curved bending member when the bending moment increases the radius. The equation for calculating radial tension was derived theoretically from mechanics as an analogy to a pressure vessel: fr where M R b d bending moment radius at member middepth member width member depth 3M 2Rbd

It was not well understood among design engineers that the above formula was applicable only to prismatic members of constant cross section. The curved pitched and tapered glulam beam was widely used, and tension perpendicular to grain separations developed in a small percentage of these beams, but enough to be of great concern to the glulam industry. Separations developed in beams where radial tension was analyzed using the above formula, and stresses were within published allowables. An extensive research program in the 1960s provided both a reduction in allowable cross-grain tension stresses for western softwoods and a revised design approach for tapered curved members. The iterative design procedure resulting from that approach is presented in the AITC Timber Construction Manual. Reinforcing perpendicular to grain, where required per the design approach for nonprismatic members, solved the problem of beam separations and has been in common use since the late 1960s. Curved glulam beams require steel reinforcing, as do reinforced-concrete beams. Two common reinforcing methods are the installation of full thread lag bolts or the placement of epoxy-embedded deformed reinforcing bars perpendicular to grain. Since radial tension stresses are a function of radius, reinforcing is required in the curved portion only. A side effect of radial reinforcing is that member dimension is locked in perpendicular to grain when it is fastened or bonded to the steel dowel. This prevents slight expansion and contraction of the timber due to moisture cycling. When the wood does shrink due to moisture loss but is restrained, checking occurs in the reinforced portion. Allowable Design Stress. Clear-wood procedures were initially used to establish allowable design stresses for dimension lumber (2- to 4-in nominal thickness),

TIMBER STRUCTURES

15.27

timbers (5-in and greater nominal thicknesses), and glulam. Beginning in the late 1960s, the glulam timber industry initiated an ongoing program of testing full-size glulam beams for stiffness and to destruction for bending stress. Here are highlights from that testing:
G

Beam strength limitation in flexure is predominately triggered by the tension face. Natural-growth characteristics, such as knots and slope of grain, have greater impact upon strength at the edges of the tension laminations in the tension zone than at the interior portion of the lamination. The older clear-wood procedure was nonconservative for the larger beam sizes. A size/volume relationship exists across the full range of member sizes such that larger members have a lower MOR than smaller members.

The following revisions in glulam design and manufacture resulted:

G

Special grades were developed for the tension face laminates with closer tolerances on natural-growth characteristics. New bending stress combinations were developed requiring the special laminate grades at the tension face. In the early 1990s a volume factor, which is a function of beam width, depth, and length, replaced the former size factor, which is a function of depth only. The volume factor reduces the allowable bending stress of larger glulam beams as compared to the former size factor.

Tabular bending and tension stresses for solid sawn dimension lumber and timbers were also adjusted. Over time there was a gradual downward trend in allowable bending and tension stresses for commonly used glulam combinations and dimension lumber grades in Douglas fir and other softwood species. Knowledge of the above state-of-the-art design revisions is imperative in evaluating older flexural or tension members. These members, originally designed in accordance with the building code in effect at the time, often do not meet current criteria. While failure is rare, many of the flexural or tension members designed to older criteria have infringed upon their safety factor. An example is a recent evaluation of glulam beam roof framing at a textile mill designed in 1959. The beam was originally designed to full allowable bending stress. Criteria changes included reduction of allowable bending stress, introduction of the volume factor, and a snow load increase from 40 to 100 lb/ft2 due to more recent snow load history. (See Case Study 13, Guilford of Maine Textile Mill.)
Loading/Overloading Dead Load. Dead load is more easily defined and calculated than the more capricious wind and seismic loadings. In forensic design review, the dead load

15.28

CHAPTER FIFTEEN

must be accurately estimated. This is relatively simple for conventional construction where the components may be seen. However, in some instances not all the construction is visible. Where poured decks of gypsum or lightweight concrete cannot be measured for depth, it may be necessary to remove portions of the structure or take a roof core. A competent roofer can remove a roof core and patch the roof, preserving the roofing bond. (See Case Study 8, Robert Frost School, and Case Study 9, Hughes Aircraft Building.) Snow and Snow Drifting. Snow loading for design review is specified by each local city, county, or state building jurisdiction. The snow load specified is usually ground snow load. Codes permit adjustment of ground snow load for roof snow load and pitch of roof dependent upon conditions. The building codes extensively cover calculation of specific snow load conditions such as drifting, sliding surcharge, overhangs, and valley design. A forensic inspection where snow load is known or suspected should consider snow weight measurement a top priority. Measuring snow depth is not adequate. The measurement should be precise and well documented. Shoveling snow into a bucket for weighing is not acceptable. Extracting a core of snow down to and including ice against the roof is the most accurate method of measurement. While specialized snow coring equipment is available, the author prefers sections of 6or 8-in stovepipe available at most hardware stores. The extracted core may be weighed wherever convenient and the snow density calculated. It is valuable to measure snow weight on both the roof and an adjacent undisturbed area of the ground. (See Case Study 12, Albertsons Grocery Warehouse.) Ponding. Water ponding on flat or nearly flat roofs may be caused by inadequate roof slope, drainage obstructions, frozen drain pipes, or deflection due to lack of stiffness in the roof framing members. Ponding problems are more common in areas where roof live loads, rather than snow loading, control the design. Under UBC the roof live load may be reduced to 12 lb/ft2 for flat roofs with tributary loaded area of 600 ft2 or more. This leaves little reserve for water ponding compared to a northern area requiring 25 lb/ft2 or more of snow load for design. Water ponding is a chicken-and-egg type of progressive-deflection problem. Weight of water accumulation causes roof beams to deflect, roof beam deflection creates a deeper pond with greater water weight, causing more deflection, causing more water weight and more deflection. This continues until balanced by beam resistance, the water escapes or the beam fails. (See Case Study 1, Port of Longview Warehouse No. 1.) Wind Load. Wind load requirements of the building code are derived from various available sources but primarily from extensive work and documents published by ASCE culminating in ASCE Standard 7-95.15 Wind is a capricious force whose maximum intensity may last for only a few seconds. Wind force is not distributed uniformly over a given area. It is a common occurrence, following a major windstorm, to find a wind-damaged structure standing adjacent to one which is intact. Most older buildings were designed without consideration for tying elements of the structure

TIMBER STRUCTURES

15.29

together, such as roof to column, column to footing, etc. An initial step in a forensic investigation of wind load damage should be wind gauge records from the National Climatic Data Center or other sources. (See Case Study 4, First Nazarene Church.) Unbalanced Load. Simple span beams do not require checking for unbalanced loading, but multiple-span and cantilever beam systems are sensitive to unbalanced loading. The model building codes all have some language requiring consideration of unbalanced loading where it creates a more severe design situation than balanced loading. The 1997 Uniform Building Code (UBC)16 in section 1607 requires unbalanced loading where such loading will result in larger members or connections. However, an exception is included in the code where alternate span loading need not be considered if the uniform roof live load is 20 lb/ft2 or more. The authors design review experience has shown that unbalanced loading is often not considered in the design of cantilever beam systems. Most, if not all, cantilever beam systems will be controlled by unbalanced live load at alternate spans, and the authors experience is counter to the above code exception which tends to sanitize cantilever beam system design without provision for unbalanced loading. (See Case Study 1, Port of Longview Warehouse No. 1.) For trusses, unbalanced loading may control web design and web-to-chord connections. Spherical dome members are primarily in compression except that unbalanced or concentrated loading may reverse them to tension.
Deterioration Decay, Termites, and Marine Borers. Under conditions favorable to their growth, fungi will attack wood, causing decay and seriously reducing the structural integrity of timber members. These organisms require four essentials to survive: food, air, moisture, and a favorable temperature. Wood decay cannot progress when the organisms lack any of these four essentials. Wood is the food source. Lacking air, continuously submerged wood will not decay from most wood-eating organisms although in saltwater it may be attacked by marine borers. The wood-eating organisms are dormant at temperatures below the middle 30F range. In normal use three of these essentials are usually present, requiring only moisture as the catalyst to induce decay. Coring and stress wave testing, as described under Methods of Testing, are useful in situ methods to determine the extent and limits of decay in structural members. (See Case Study 7, Hayward Field Grandstand.) Termite activity is a function of temperature, being more prevalent in warmer climates such as Florida, with reduced activity as temperature decreases. Termites are virtually nonexistent in the northern heartland states away from the tempering effect of oceans. Subterranean termites maintain ground colonies with mud tunnels to the wood structure above. Wood exposed to saltwater environments is subject to a family of wormlike marine borers which burrow below the wood surface. As with termites, their

15.30

CHAPTER FIFTEEN

activity is a function of temperature, and unprotected woods in tropical waters have become nonfunctional due to marine borers within a period of a few months. Chemical Deterioration. Wood is considered superior to many other materials of construction in resistance to chemical action but is not totally free from chemical deterioration. While working in the wet environment of the Aleutian Islands, the author saw untreated timber bridge members in good condition where the steel bolts had disintegrated to rust traces due to corrosion caused by sea salt. He once designed a timber domed roof covering a chemical coking operation so corrosive that the previous steel roof had collapsed after a reasonably short period of service. Wood structures are used for bulk storage of chemicals such as urea fertilizer which corrodes unprotected steel rapidly. The heartwood of many species is more resistant to chemical action than sapwood. Water and some organic liquids such as alcohol cause wood to soften and swell, causing a temporary strength reduction. Acids and acid salts cause permanent strength loss due to chemical changes in the cell structure. Alkalines cause permanent strength loss due to destruction of lignins which bind wood fibers together. Heat. Wood strength decreases with higher temperature and increases with lower temperature. Tabular design values are set at approximately 20C. Temporary strength reduction occurs at temperatures to approximately 150F, and permanent strength reduction occurs beyond that level. (See paragraph entitled Temperature.)
Lateral Buckling

Columns, arches, and truss members stressed in compression are usually controlled by buckling or the interaction of buckling and compression, rather than by pure compression. Trusses and arches are rarely stable alone until secured into a diaphragm or some type of lateral bracing system; therefore buckling problems with these structural systems are most likely to occur during construction. Inadequate bracing has been the primary cause of truss problems that the author has investigated. Causes include inadequate temporary bracing, diaphragm construction lagging behind truss or arch placement, and pressure to release expensive rented cranes. (See Case Study 2, McMinnville Town Center; Case Study 3, Lighter than Air Hangars; and Case Study 5, Rosemont Horizon Arena.)
Notching or Tapering Tension Face

Notching or tapering the tension face of a stressed timber member severs stressed wood fibers and sets up a potential distress situation that is difficult to analyze accurately. A good rule of thumb is to never notch or taper the tension side of a highly stressed timber member. (See Case Study 4, First Nazarene Church, and paragraph entitled Beams.)

TIMBER STRUCTURES

15.31

Eccentricity

Older bowstring trusses were often constructed using double chords with single webs in the plane between chords. In this case the webs could not be connected concentrically because of web-to-web interference. Bowstring trusses with small web forces were usually able to tolerate this eccentricity. Older precomputer structural analyses by graphic analysis or the method of joints assumed pinned concentric joints and ignored web-to-chord eccentricity. Truss heel eccentricity where centroid intersection of heel members is eccentric with the truss reaction induces a torque into the heel connection that may create distress in trusses not designed for this condition. Internal truss member eccentricity where centroid intersections of adjacent web members are eccentric with chords creates the same condition. For bowstring trusses, distress due to heel eccentricity is more likely than distress due to internal web eccentricity as the member forces at the heel are much higher. (See Fig. 15.6 and Case Study 6, Rainier Beach Library, and Case Study 10, Sheridan High School Gym.)
Secondary Effects

Secondary effects and particularly the effect of nodal rotation on trusses should be considered in design or forensic evaluation. Ideally, web-to-chord truss joints should be as designedi.e., welded if designed fixed, or frictionless pins if designed pinned. Most web connections in timber trusses are designed as pinned, but have partial fixity due to restraints in the connection. If long, rigid steel side plates, designed to transfer axial loads only, are used for web-to-chord connections and the truss is flexible with considerable nodal rotation, then the connectors will resist that rotation by creation of moment couples in the web members. In severe cases these couples create a cross-grain tension force which can split the truss member. See Case Study 6, Rainier Beach Library, in which the outer truss web split due to a combination of nodal rotation and truss heel eccentricity. See Case Study 15, Hood River Valley High School Theater, where nodal rotation at a wood chord-wood web truss with glued connections failed due to cross-grain tension stresses induced at the web-to-chord connections.
State-of-the-Art Design Changes

The state of the art in timber design has changed substantially over time, but many structures the forensic engineer is retained to evaluate are older and designed to a less rigorous code standard. The safety factor is usually adequate to keep older structures viable, but effective reduction of the safety factor due to evolution of design criteria leaves these structures less able to carry additional or unexpected loading. Understanding state-of-the-art changes in tabular stresses and design procedure during the life of the structure being evaluated is vital to a forensic evalu-

15.32

CHAPTER FIFTEEN

e e

e e

TRUSS A
12Illustrating bowstring truss where centroid intersection of heel members are eccentric with truss reaction. Illustrating bowstring truss where centroid intersection of adjacent web members are eccentric with chords.

TRUSS B
Illustrating bowstring truss where centroid intersection of adjacent web members are concentric with chords and where centroid intersection of heel members are concentric with truss reaction.
FIGURE 15.6

Truss member eccentricity.

ation. (See paragraph entitled Evolution of Engineered Timber Design and Case Study 1, Port of Longview Warehouse No. 1; Case Study 9, Hughes Aircraft Building; Case Study 10, Sheridan High School Gym; Case Study 11, Lucky Strike Lanes; and Case Study 13, Guilford of Maine Textile Mill.)
Compatibility of Analysis and Structural Performance

All structural analyses involve assumptions such as fixed or pinned, support conditions, etc. Analyses also involve simplifications on account of constraints of time and economics. Most of the assumptions and simplifications are conservative. In lateral analysis the engineer may use only those stiffening elements for which design values are available. Lateral bracing is analyzed on contributions

TIMBER STRUCTURES

15.33

from shear walls, diaphragms, cross-bracing, rigid frames, etc. Not considered are contributions from interior partitions, friction, and other structural and nonstructural elements which contribute to bracing. Timber dome analysis usually ignores the shell action of roof sheathing. The fact that structures may perform somewhat differently than the analysis assumptions may not always be on the conservative side and could be a factor in nonperformance. An old engineering expression says, Structures are smarter than the engineers who design them, meaning structures find their own way to function. In reviewing nonperformance, one possibility to keep in mind is that the structure is performing in a manner different from the assumptions used in the analysis.

STRUCTURAL EVALUATION OF IN-SITU TIMBER STRUCTURES Forensic structural evaluation of timber structures in place requires a strong engineering background and judgment tempered by experience. It is elementary that the physical data relating to a member or structure in question must be documented in regard to specie, grade, age, loading history, dimensions, site conditions, wood moisture content, member condition, etc. Some of this may be found in archives of the owner or the building department. The remainder must be determined on site or from documents. This may involve inclement weather, unpleasant site conditions, and inconvenient time frames. Avoid undocumented second-hand information. Structures do not become distressed or fail without a reason. A forensic engineer must identify and document in a report to his or her client the reasons for a structural problem. Knowledge of the subject matter, in particular engineering properties of wood and timber characteristics, is necessary to evaluate structural timber members. Access to older, out-of-date codes and standards is of great value as evaluation often involves older structures. All possible in situ plus laboratory testing within budget constraints should be performed and guesswork minimized. Nondestructive tests such as wood moisture content readings and stress wave testing are possible on most structures. Where specimens are available, bending, strip tension, compression, and MOE tests from a laboratory are invaluable. Test data enhance a forensic report by reducing speculation in the conclusions. Full-size load testing of bridges and buildings, if done at all, should be used with caution because of the problems of isolating the variables and possible damage to adjacent members or nonstructural elements. It is difficult to know what percentage of the design load may be applied without damage to some of the members or connections. The tenacity and inquiring mind of a detective are a great asset for a forensic engineer. Keep an open mind and be prepared to learn something from every commission. Expect the unexpected. (See Case Study 14, Church of Christ.)

15.34

CHAPTER FIFTEEN

TEMPORARY AND PERMANENT REMEDIAL REPAIRS Simply stated, any structural repair must satisfy the results of the structural analysis. The repair of members and connections must satisfy the shear, bending moment, tension compression, etc., from the analysis plus any condition such as eccentricity introduced by the repair. As such, the repair engineer must have analysis data available prior to final repair design. When a condition of structural distress is first identified, the situation is usually a state of emergency. Shoring may be required to prevent collapse before there is time for detailed analysis. Emergency shoring usually consists of what is readily available. Wood members have an advantage in that they may be easily field-trimmed to length. Round wood poles and square or rectangular timbers work well. If only dimension lumber is available, it may be nailed into T, H, or box-shaped members to resist buckling. Wood or steel wedges may be used to bring shoring tight under a distressed member. Engineered tubular steel shoring with known load capacity is ideal, if available. When trusses from the lower chord are shored, it is usually necessary to shore all panel points. This is particularly true at bowstring trusses where web-to-chord connections are relatively small due to axial web forces which are small compared to axial chord forces. The shore becomes a truss reaction, and webs fastening to the lower chord close to the shore tend to act as the top chord of a reconfigured truss. Partial shoring of a bowstring truss will cause substantial increase in web compression adjacent to the shore and may cause web-to-chord failure and possible truss collapse. Shoring under the top chord of a distressed truss is preferable to lower chord shoring but more difficult to accomplish. In a recent truss failure investigation by the author, the analysis showed that placing a crosswall tight under a bowstring truss lower chord increased compression at one web from 3 to 78 kips and caused progressive web failure and subsequently truss chord failure. In the case of overloaded beams, placing additional columns to shorten the span may be the most favorable solution. (See Case Study 13, Guilford of Maine Textile Mill.)
Shoring. Replacement. Replacement of a distressed member inside a structure is usually not feasible owing to adjacent members and interference with other building elements. In those situations where replacement is possible, it is usually more economical to replace a member than to design and install a repair, as most repairs are labor-intensive. Replacement of deteriorated truss members is feasible where the full truss may be stress-relieved by top chord shoring.

Addition of a new structural member where the original member is physically left in place is often a solution for reinforcing a distressed member in an industrial structure where appearance is not crucial. The new sistering member
Sistering.

TIMBER STRUCTURES

15.35

should be designed to carry all the load and should be fastened so as to transfer forces from the original members if required. (See Case Study 3, Lighter than Air Hangars.) Sistering members of like materials are preferred for ease of connection, but steel channel or plate may be used for sistering timber members. Sawn timber structures fastened with bolts and shear plates or split rings should have bolts tightened after moisture equilibrium is achieved, and periodically if bolts loosen. Large sawn timbers are too large to be kiln-dried and are usually installed wet with shrinkage occurring after installation. Shrinkage at connections fastened with shear plates or split-ring timber connectors may cause a separation between timbers at the load transfer point, which reduces the bearing area of wood against the plate or ring, thereby reducing connector capacity. Glulam is kiln-dried and more dimensionally stable, so it is less in need of periodic bolt tightening.
Bolt Tightening.

Clamps and stitch bolts are particularly useful to prevent propagation of timber separations in truss chords and webs. Usually a stitch bolt of 1 2-in diameter with washers is adequate to close a separation. Stitch bolts should only be used at stressed members if the loss of section by the additional bolthole may be tolerated. A vertical stitch bolt adjacent to truss lower chord-web connections may be used to resist the cross-grain tension stress created by ceiling loading on the lower chord below the connection. Clamps consisting of steel plate, angle, or channel above and below a member and fastened by bolts each side of the member may be used to close separations or resist cross-grain tension forces without loss of section.
Clamps and Stitch Bolts. Beam Stretcher Moment Connections.

A flexural member may be spliced providing that the connection accounts for all the shear, axial force, and bending moment felt by the member. Moment splice connections are used in original designs in glulam arches to shorten the member for shipment. They are also used at knees of glulam rigid frames. Moment connections may also be used in repair to replace a portion of a flexural member. Decayed or damaged arch legs may be repaired by removing the damaged portion and fitting a new moment-connected base shoe. (See Case Study 14, Church of Christ.) In a recent investigation by the author at a papermill, a glulam beam end supported at a column had decayed in the wet environment. The decayed ends were removed and fitted with steel moment connected caps which fit the original beam seat. The beam was salvaged without full beam replacement.

Post Tensioning.

Post tensioning of beams involves application of an external compression force into the member, preferably at a location causing bending moments which counter those due to the supported loads. Because this force is usually applied by a steel rod or cable strand in tension, it is termed post tensioning. Numerous tensioning strand configurations are possible, the most common

15.36

CHAPTER FIFTEEN

being a tension strand at or near the beam soffit to create maximum negative moment due to the lever arm distance between the tension strand and the beam neutral axis. Tension strands below the beam soffit may use a vertical spreader member to increase the tension strand lever arm, creating a sort of king post truss. Since strand tension may be quite high, anchorage of strand to beam is critical. Mild steel shapes, mild steel threaded rod, and cable have been used successfully as tension strands. Turnbuckles may be used to apply tension to steel rod tension strands. Cable of the type used for prestressed concrete has an advantage in that inexpensive button type of end anchors may be fastened to the cable and a predetermined amount of tension applied using a calibrated hydraulic jack. The author prefers this system over use of threaded rods for tensioning, as the amount of force applied may be accurately controlled without risk of overtensioning. Post tensioning of trusses may be utilized to counteract overloading or calculated overstress due to changes of criteria. Post tensioning of trusses usually consists of a tension strand at middepth of the lower chord applying an axial force to the chord without inducing moment. Truss post tensioning will reduce lower chord tension and reduce the possibility of lower chord distress.
Radial Tension/Cross-Grain Tension Reinforcement. Wood separations due to radial tension induced by bending of curved glulam members are rarely seen today as radial reinforcing has become a typical design procedure installed during manufacture. Field repairs of unreinforced beams containing radial tension separations may be engineered using a lag bolt field installed to carry the tension perpendicular to grain force. The beam section must be adequate to sustain section loss from the repair dowel. To close the separation, the lag bolt may be installed from below in holes 1 16 in oversize below the separation and above the separation in undersize lead holes as prescribed in the NDS for lag bolts in withdrawal for the wood specie present. Stitch bolts perpendicular to grain may be used to resist tension perpendicular to grain forces in ledgers resisting lateral loads and at truss lower chords at web connections.

Load-induced shear separations are rare in wood as a rather large safety factor separates actual ultimate shear strength and tabular design stresses in horizontal shear. If delamination occurs in glulam, reinforcing against a potential shear failure may be appropriate. The potential failure mode for this situation is separation of the beam into two segments along a horizontal plane and subsequent flexural failure due to reduced section modulus. If a separation parallel to grain due to delamination, checking, etc., has the potential for shear failure of the member, then shear reinforcing is appropriate. The reinforcing must transfer shear between the separated segments, which may be accomplished by embedded dowels or exterior side plates. The dowel shear transfer or side plate connectors must be engineered to transfer the calculated shear. Lag bolts or epoxy-embedded dowels, placed perpendicular to grain, may be used with the
Shear Reinforcement.

TIMBER STRUCTURES

15.37

added advantage of the dowels being able to resist any perpendicular to grain forces. A precise evaluation of beam net section containing reinforcing dowels would be with a transformed section using full section for the compression side of the neutral axis, and the full section less dowel area removed for the tension side of the neutral axis. A quicker, more conservative assumption would be the full dowel area removed from the entire net section. Data sufficient for an engineer to design either a lag or epoxy-embedded dowel reinforcing are covered in AITC 404-92, which is an appendix to Inspection Manual AITC 200-92. Steel side plates on each side of a beam connected with timber rivets may also be used to transfer shear between separated wood segments. The total shear transfer required may be calculated and the repair designed per the procedure outlined in the NDS.

EXAMPLES AND CASE STUDIES Several case studies are included to illustrate examples of distress, forensic evaluation, and repairs discussed in the body of this chapter. All the case studies included here were forensic investigations or evaluations conducted by the authors firm.
Case Study 1. Port of Longview Warehouse No. 1 Longview, Washington

This case study illustrates

G G G

Ponding Unbalanced loading at cantilevered beams State-of-the-art criteria revisions

A glulam framed cantilevered beam warehouse roof framing system was designed in 1965 per the following criteria: Glulam combination A with allowable bending stress of 2600 lb/in2, 20 lb/ft2 roof snow load, and 12 lb/ft2 roof dead load. Using balanced loading for the full beam length, including 16-ft overhangs on each side, the beams were acceptable for shear and bending per the original design criteria, and deflection was within L/240. The authors firm was retained for forensic evaluation in 1973 following collapse of a portion of the roof after a heavy rainstorm. See Fig. 15.7. Several factors contributed to the overstressed condition leading to collapse, some of which alone were of relatively minor significance. A 20 lb/ft2 roof snow load was permitted by the building code in 1965, but by 1970 it had been raised to 25 lb/ft2. The design dead load of 12 lb/ft2 was about 1 lb/ft2 low, which is about the weight of the sprinkler system. By 1973 the glulam A combination at

15.38

CHAPTER FIFTEEN

FIGURE 15.7

Collapse of cantilever beam system due to water ponding. Port of Longview Warehouse No. 1, Longview, Washington.

Fb 2600 lb/in2 had been discontinued, and the 24F combination was widely used. Based upon present criteria, the A combination should be used at approximately 2200 lb/in2 for an appropriate safety factor. A plan sketch of roof framing in Fig. 15.8 shows that the four building corners were notched in plan view; i.e., overhangs were not present at the corners of the building. This caused the overhang portion of the end beams to carry only one-half the roof load of comparable interior beams. Reduction of roof load at overhangs caused bending moments in the end beams to be higher than in interior beams. The beams appear to have been sized for the less severe case. The beams appear to have been sized without consideration for unbalanced loading. Cantilevered beam systems are relatively flexible because of reduced beam moments due to cantilever action and the resulting smaller beam size. The roof was constructed flat or close to flat. Bracing of beams in the negative moment portion was inadequate but did not appear to contribute to the collapse. Calculations by the author showed the following: 4,100,000 in lb maximum moment with unbalanced snow load and reduced tributary area at roof overhang portion for end beams per 1995 criteria maximum moment with balanced snow load and no consideration for reduced tributary area at roof overhang portion per 1965 criteria

2,850,000 in lb

TIMBER STRUCTURES

15.39

COLLAPSED

11' 16' 70' 70' 16'

FIGURE 15.8 Warehouse roof plan. Port of Longview Warehouse No. 1,

Longview, Washington.

9-in

275 8-in glulam beams:

247.5 in2

1134 in3

The original 1965 design condition using 1965 design criteria and use of balanced snow load (SL) plus dead loading at beam overhang yields fb fb 2,850,000 in lb 1134 in3 2513 lb/in2 2571 lb/in2

(2600 lb/in2)(1.15)(0.86) CD C F 2513 lb/in2 2571 lb/in2 0.98

fb Fb

Maximum SL deflection

L 262

15.40

CHAPTER FIFTEEN

Use of 1965 design criteria with unbalanced snow load and reduced tributary area at roof overhang yields: fb fb Fb 4,100,000 in lb 1134 in3 3616 lb/in2 2571 lb/in2 1.41 L 210 3616 lb/in2

Maximum SL deflection

Use of 1973 design criteria with unbalanced snow load and reduced tributary area at roof overhang yields: fb Fb 4,100,000 in lb 1134 in3 38 lb/ft 2 32 lb/ft 2 4293 lb/in2 2324 lb/in2

(2400 lb/in2)(1.15)(0.86) CD CF 4293 lb/in2 2324 lb/in2 1.81

fb Fb

Use of 1998 design criteria with unbalanced snow load and reduced tributary area at roof overhang yields: CV Fb 0.746 1880 lb/in2

(2200 lb/in2)(1.15)(0.746) CD CV fb Fb 4293 lb/in2 1880 lb/in2

2.28

In this case the beam was initially undersized for good design practice and had too many strikes against it with water ponding creating the failure trigger. The actual water loading at failure is unknown, but the above calculations show the beam substantially overstressed without ponding considerations. Examination of the failed beam showed classical flexural failure of a properly manufactured beam.
Conclusion. Case Study 2. McMinnville Town Center McMinnville, Oregon

This case study illustrates lateral buckling.

TIMBER STRUCTURES

15.41

FIGURE 15.9

Collapse of light timber trusses during construction. McMinnville Town Center, McMinnville, Oregon. (Photo courtesy of McMinnville News Register.)

A series of one-story wood-framed buildings forming a retail sales strip mall was under construction in 1977. Proprietary wood chord-wood web, parallel chord roof trusses 42 in deep were used with 32-in truss spacing and 68-ft truss span. Truss collapse had occurred at a portion of the complex while some trusses remained standing. See Fig. 15.9. The forensic evaluation was conducted in June and July of 1977. The position of the collapsed trusses on the ground indicated top chord buckling and truss rollover. No bracing was apparent. No detailed structural analysis was performed by the author. The problem was not a strength deficiency, as the trusses at time of collapse were carrying approximately 7 percent of their total design load. The truss supplier had designed and furnished bracing elements consisting of bridging and plywood shear panels between trusses. None of this bracing or the roof diaphragm had been installed. All the trusses were installed by the contractor, who was planning to add bracing later in order to save crane time. The forensic evaluation determined the collapse was caused solely by lateral buckling of truss top chords due to inadequate bracing. Litigation between the contractor and truss supplier was decided in favor of the truss supplier. The structure was rebuilt with identical, properly braced trusses.
Case Study 3. Lighter than Air Hangars Moffett Field Naval Air Station Moffett Field, California

This case study illustrates lateral buckling.

15.42

CHAPTER FIFTEEN

During the early portion of World War II, Germany waged a devastating sea war in the Atlantic, using its U-boats against Allied shipping. To counter this, the U.S. Navy was charged with developing a lighter-than-air (LTA) blimp patrol program, as the blimp could stay aloft for extended periods, spot submarines, and escort ship convoys. To house the blimp squadrons, a total of 17 LTA hangars were constructed of fire-retardant treated solid sawn timbers. Two of these hangars were constructed at Moffett Field, California. Among the largest structures ever built of timber, the hangars are 1040 ft in length. The two-hinged trussed frames span 296 ft 6 in and rise to 170 ft 4 in in height. Among the problems identified in the forensic evaluation were 36 buckled frame chords. The immense height placed some restrictions upon the repair procedures considered. The frames are constructed of double chords, double diagonal webs, and single vertical webs. The repair procedure selected consisted of placing solid glulam blocking between the double buckled chords with glulam strong-back sistering members on each side to realign buckled chords. The repair procedure was simple in concept, but more complex in execution. Buckling of frame chords was a problem of the original design in selecting chord sizes of inadequate thickness or l/d ratio. About one-half of the 17 hangars constructed utilized 4-in nominal chords, and the remainder used 3-in nominal chords. Little or no buckling occurred at the 4-in nominal chords. Buckled chords at the Moffett Field hangar were confined to the 3-in nominal chords. A more detailed description of the forensic evaluation is contained in Proceedings of the ASCE Structures Congress, New Orleans, LA, 1986.17 The lower edges of frame lower chords have been covered with nailed blocking to discourage pigeon nesting. Figure 15.10 shows buckled lower chords with

FIGURE 15.10

Repair of timber frame lower chords at LTA hangar. Moffett Field Naval Air Station, Moffett Field, California. (a) Buckled frames with bird blocking present. (b) Buckled frames with bird blocking removed. (c) Repaired frames.

TIMBER STRUCTURES

15.43

bird blocking in place, buckled lower chords with bird blocking removed, and repaired lower chords. Following closure of the base by the Navy, the hangars were transferred to NASA. The 1989 Loma Prieta earthquake caused extensive damage to one of the hangars, and the authors firm was retained by NASA for forensic evaluation and development of structural repairs relating to the seismic damage.
Case Study 4. First Nazarene Church Portland, Oregon

This case study illustrates notching and tapering of the tension face of the beam. The roof structure of a large church was designed using glulam roof framing. Two beams 141 4 in 881 2 in spanning 118 ft supported the entire roof system with glulam beam and light truss subframing. To satisfy architectural constraints the tension faces of beams were tapered 30 ft at each end. To facilitate the upper column seat, a notch was located within the upper taper cut. See roof framing plan and beam elevation sketches in Figs. 15.11 and 15.12.

121'-0"

FIGURE 15.11

Roof framing plan. First Nazarene Church, Portland, Oregon.

121'-0"

15.44

CHAPTER FIFTEEN 12 19/16 881/2" FAILURE LINE

141/4" GLULAM

10'-0 5/8" 30'-0" 68'-0 5/8" 128'-0 5/8" 118'-0" c - c COLUMNS L L 30'-0"

FIGURE 15.12

Roof beam elevation. First Nazarene Church, Portland, Oregon.

FIGURE 15.13 Collapsed structure. First Nazarene Church, Portland, Oregon.

TIMBER STRUCTURES

15.45

The church had been occupied approximately one year when total collapse of the roof occurred on September 9, 1981. Approximately 20 min elapsed between the time failure was recognized and the time when total collapse occurred. Sticking doors and other building deformations had been occurring for a period of weeks or months preceding collapse, but were not initially recognized as a failure in progress. The authors firm was retained for forensic evaluation by the beam supplier. See Fig. 15.13. The collapse occurred under mild weather conditions with no snow or other roof live loads present. The collapse was triggered by a separation at the notch located in the upper taper cut near the beam middepth. The separation progressed toward the midspan of the beam, separating the beam into two segments.

FIGURE 15.14 Beam following recovery from collapse. First Nazarene Church, Portland, Oregon.

15.46

CHAPTER FIFTEEN

Flexural failure occurred when bending stress exceeded resistance at the top portion of the separated beam. The two beams failed in identical fashion. See Fig. 15.14. Forensic evaluation determined that failure was solely due to effects of the notch and taper cut. The beam was otherwise sized in accordance with code requirements. The church suffered substantial additional damage when a windstorm collapsed some of the remaining unbraced walls. Broken water pipes from the collapsed walls caused additional damage to what remained of the building. The church was rebuilt to the same configuration, using moment connected steel girders to replace the two failed beams. With the structure in place, a replacement timber beam was too heavy to lift by crane.

Case Study 5. Rosemont Horizon Arena Rosemont, Illinois

This case study illustrates lateral buckling. The sports and general-purpose arena was constructed in 1979 adjacent to Chicagos OHare Field. A total of 14 glulam barrel arches, spaced at 24-ft centers and spanning approximately 288 ft, frame the sports arena structure. The building footprint is approximately 288 ft 382 ft in plan. Arches are 103 4 in 731 2 in 24F glulam. Radius at top of barrel arch is 431 ft 8 in. The glulam arches were supplied in three segments, connected by two moment connections. Framing between arches consisted of glulam purlins at 4-ft centers with wood decking/diaphragm over. On August 13, 1979, at a point when 12 of the 14 arches were in place, collapse occurred rapidly, killing five workers and injuring 19 others. Initially, several collapse theories were discussed in the extensive media coverage including vibration and wingtip vortex from low-flying aircraft. Placement of purlins and decking/diaphragm was following well behind erection of the arches. Causes were attributed to lack of workers available, lack of coordination between the two operations, and schedule pressure. Every sixth purlin at 24-ft centers was of a larger size and designed as a tie beam bolted in place between arches. Investigation revealed that about one-third of these bolts were in place, and no bolts secured the last arch set. Compressive buckling of the unrestrained arch, due to inadequate bracing, is considered to be the sole cause of failure. Arches ahead of the decking/diaphragm placement had inadequate or nonexistent lateral bracing. The rapid collapse also suggests buckling, whereas flexural failure is usually accompanied by noise and incremental movements over a period of time. The arena was rebuilt per the original design.

TIMBER STRUCTURES

15.47

Case Study 6. Rainier Beach Library Seattle, Washington

This case study illustrates

G G

Nodal rotation of fixed truss connections Secondary effects

y ; y ;
5'-0"

A branch of the Seattle Public Library system was constructed in 1980, framed with timber roof trusses. Parallel chord trusses were designed using glulam chords and webs and connected with bolted steel plates. Truss span is 46 ft, and overall truss depth is 5 ft. See Fig. 15.15. Chords are continuous, and web-to-chord con-

L = 46 FT

y ;

FIGURE 15.15

Truss profile. Rainier Beach Library, Seattle, Washington.

7'-8"
28

4" 4" 4" 4" 4"

"

4" 3" 3" 4" 51/8' x 9" GLULAM

4"

c COLUMN L

51 8 GL / ' x UL 6" AM

51/8' x 6" GLULAM

4"

4"

FAILURE PLANE

4"

81/2

4" 51/8' x 71/2" GLULAM

4"

4"

4"

1'21 / 2"

FIGURE 15.16 Truss heel detail. Failure occurred at outer diagonal web. Rainier Beach Library, Seattle, Washington.

73/4

5 FT

15.48

CHAPTER FIFTEEN

nections were analyzed as pinned. Member sizing and bolt capacity were adequate for the primary loading condition. The web-to-chord connections consisted of welded 3-in 1 4-in steel fin plates on each side of truss following the profile of truss members. All connectors were 1-in bolts. The welded truss connection plates created partial fixity at truss joints. When the truss deflected under loading and nodal joints rotated, the multiple bolts in a row resisted that rotation by creation of moment couples in the glulam truss members, inducing a force perpendicular to grain and thereby stressing the glulam member in cross-grain tension. (See Fig. 15.16.) In 1985, one outer diagonal web separated during a snow loading condition due to the moment couple induced by truss joint rotation. At the truss heel the distance from intersection of top chord and outer bolt of web connection is 28 in. The truss was shored and the split web replaced, including stitch bolts to resist perpendicular to grain forces. The web failure was caused by a combination of nodal rotation and torque induced from heel eccentricity causing cross-grain tension in web members. (See Figs. 15.6 and 15.16.) This is an example of an adequate design for primary forces, but failure of a truss member caused by secondary effects of joint rotation as the truss deflected under loading. The moment couple induced in truss members was high due to a relatively flexible truss and a long moment arm caused by the rigid steel plate connection.

Case Study 7. Hayward Field West Grandstand University of Oregon Eugene, Oregon

This case study illustrates decay. The University of Oregon has traditionally fielded a strong track and field program, and the Hayward Field facility is the heart of that program. The west grandstand, framed in glulam timber, was constructed in 1974. Column-supported cantilevered glulam frames carry both the roof and the grandstand seating. The grandstand contains 11 glulam frames. The outer frame at each end is preservative-treated with pentachlorophenol in light solvent, and the inner 9 frames are untreated. Glulam beams spanning between the primary frames form treads and risers supporting the grandstand seating. These treads and risers restrain but do not prevent passage of free water from the grandstand to the glulam frames below. Free water is present at the front portion of the grandstand from wind-blown rain and throughout the grandstand from power washing the underside of the seats following an event in the arena. Small amounts of moisture escaping through the grandstand wetted the glulam frame below and tended to accumulate in the steel base shoe supporting the frame. The problem was not recognized until substantial decay damage had occurred. A forensic evaluation was conducted, examining the lower leg of all 11 frames in the spring of 1998, using nondestructive stress wave testing. The outer frames

FIGURE 15.17

Mapping limits of decay at glulam frame using stresswave testing. Hayward Field Grandstand -University of Oregon, Eugene, Oregon.

15.49

15.50

CHAPTER FIFTEEN

containing preservative treatment were in nearly new condition. The nine untreated interior frames were in various stages of decay. Stress wave testing located the limits of decay, which were mapped to scale. Figure 15.17 illustrates stress wave mapping of decay at one of the glulam frames. Repair design is under way by the owners structural engineer.

Case Study 8. Robert Frost School Arlington Heights, Illinois

This case study illustrates overloading. The school classroom and gymnasium roof structure is framed with glulam roof beams. The original building dates to 1964 for which southern pine glulam was furnished. A major addition in 1966 utilized Douglas fir glulam. In February 1990, the authors firm was retained to conduct a forensic evaluation of distressed glulam roof beams at several classrooms. Six glulam beams were found to have failure of the tension lamination at or adjacent to the end joint. Those beams were shored, and no roof collapse occurred. All the beams appeared to be deflected in excess of what would be expected for the conditions of span, beam size, and apparent loading. Stress wave testing was conducted to measure MOE, and cores were extracted to check material density and shear-tested for face bonding adhesive strength. End joints were examined carefully. The material stiffness was normal, and the adhesive strength and wood condition were good, except for the distress locations. Most beams had lost their initial camber and showed residual deflection which, in the absence of significant snow load, suggested long-term heavy dead load. The original roof deck design consisted of 2-in poured gypsum on 1 2-in gypsum formboard on steel bulb tees with 1-in rigid insulation, built-up roofing, and gravel. The school had recently been reroofed, and the roof level modified to improve roof drainage. Original gravel, roofing, and insulation were removed, and over the original gypsum deck was placed a leveling coat of insulating concrete, 1-in styrene insulation, and new roofing. Our firm was issued a letter indicating 20 lb/ft2 total roof dead load including beams. Original design data were not available, but calculations showed the combination B glulam beams (Fb 2400 lb/in2) were designed to carry approximately this amount of loading with code-prescribed 30 lb/ft 2 snow loading. Due to the uncertainty of calculating dead load with tapered elements or this type of construction, roof cores were requested. The request was initially rejected due to constraints of time, cost, and reluctance to bore holes in a new roof. The company installing the reroof was eventually retained for roof coring, but these cores were of roofing only, indicating a roofing weight of approximately 17 lb/ft 2. The cores did not penetrate the gypsum decking. A request was made for cores through the full roof and decking. Later an independent test laboratory was retained for coring of the full roof and decking. Roof weight varied due to tapering of the original gypsum decking and tapering of the

TIMBER STRUCTURES

15.51

reroof to achieve drainage, but a design dead load of 55 lb/ft 2 was recommended by the owners structural engineer based upon the core data. Beams designed for 20 to 25 lb/ft 2 dead load had initial dead load in excess of this amount due to gypsum deck thicker than specified. The reroof added more dead load to 55 lb/ft 2. The glulam beams performed exceptionally well under 25 years of substantial overload. Steel flitch plate beam reinforcing was designed by the owners structural engineer.
Case Study 9. Hughes Aircraft Building El Segundo, California

This case study illustrates

G G G

Overloading War production directive State-of-the-art criteria revisions

The Hughes R-5 building covers approximately 11 acres, and the roof framing consists of approximately 540 timber bowstring trusses spanning 40 ft and spaced at 20-ft centers. The building was constructed during World War II and is located on Nash Street adjacent to the Los Angeles International Airport. At one time it was used as an assembly facility for Nash automobiles.

FIGURE 15.18

Roof mounted HVAC equipment. Hughes Aircraft Company, El Segundo,

California.

15.52

CHAPTER FIFTEEN

The authors firm was retained in 1992 to evaluate approximately 20 percent of the roof structure following several truss failures. The bowstring trusses of solid sawn timber showed evidence of considerable weathering. A design review indicated the trusses were probably designed for 7 lb/ft2 roof dead load, no ceiling load, and 12 lb/ft2 live load using elevated allowable stresses per the War Production Directive. A variety of patching, splicing, and repair work showed evidence of past truss distress. The top chord was not fastened directly to a diaphragm and was stabilized only by the continuity of vertical truss webs to the roof above. An analysis and design check showed the truss adequate for 9.9 lb/ft2 dead load and no live load per the 1991 UBC. Although the trusses were initially marginal and well overstressed by current criteria, it is probable they would have performed without distress in the absence of roof overloading. The failure trigger for the distress observed was substantial overloading from roof-mounted HVAC equipment, as shown in Fig. 15.18. The roof surface is a maze of pipes, walkways, and mechanical equipment, which includes chilled water lines, ductwork, cooling and ventilating units, walkways, and platforms. The mechanical equipment loading created an extreme situation of roof overloading. From a structural standpoint, the simplest repair option was a new beam under each distressed truss, as the span was only 40 ft. The client elected to utilize new steel sistering trusses fastened on each side of the distressed original trusses.
Case Study 10. Sheridan High School Gym Sheridan, Oregon

This case study illustrates

G G

Truss heel eccentricity State-of-the-art criteria revisions

The high school gym was constructed in 1949 using sawn timber bowstring roof trusses spanning 68 ft and spaced at 20-ft centers. A review of the truss suppliers documents indicates the trusses were designed to carry 640 lb/ft total dead plus snow load per 1949 design criteria. When investigated, the trusses were carrying dead load which added to the code-prescribed snow load amounting to 1040 lb/ft. The authors firm was retained in 1994 to evaluate the trusses following extensive splitting at both top and lower chords near the vicinity of truss heels at several trusses. An analysis and design check showed the lower chord in tension to be 85 percent overstressed by present criteria. The overstress resulted from state-ofthe-art criteria revisions over time, and additional ceiling dead load beyond that considered in the original design. The failure trigger for splitting and distress of chords at truss heels was heel eccentricity of approximately 6 in. It is not known whether truss heel eccentricity resulted from not considering heel eccentricity significant initially or whether the columns were set too far apart. Heel eccentricity creates a torque induced into the heel joint which is the product of the eccentricity and the support reaction. This style

TIMBER STRUCTURES

15.53

of truss was not designed to resist heel eccentricity, and for full design load at the east truss heel, torque was nearly one-quarter million inch-pounds. (See Fig. 15.6.) The factors contributing to the distress at truss heels were added ceiling load, state-of-the-art criteria changes, and heel eccentricity, with heel eccentricity having the greatest impact upon the distressed truss heels. The school district elected to demolish the building rather than repair the trusses.

Case Study 11. Lucky Strike Lanes Tucson, Arizona

This case study illustrates

G G

Cross-grain tension at web-lower chord connection State-of-the-art criteria revisions

The bowling alley structure was constructed in about 1954. The roof is framed with solid sawn timber bowstring monochord trusses with a span of 120 ft and spacing of 19 ft 6 in. Ceiling framing supported by the truss lower chord carries HVAC equipment and the finish ceiling. The authors firm was retained in 1995 to evaluate separations in the truss lower chords. Of the seven trusses inspected, one contained an old lower chord separation, subsequently repaired, and three others showed unrepaired lower chord separations. The most severe location of distress amounted to a full severing of the lower chord, and that truss was shored. The cause of lower chord separations was cross-grain tension due to the component of web members at the chord connection acting perpendicular to grain at the truss chord and the weight of the ceiling acting downward. Truss analysis assumes members to be line or bar elements and connections to be point nodes. The actual lower chord is an 8 12 timber member with the web connection near the top edge of chord and ceiling supported by framing near the lower edge of the chord. The ceiling load created cross-grain tension stress acting continuously upon the chord over its service life and eventually pulled the lower chord apart perpendicular to grain. A simple clamp or stitch bolt on each side of each web-to-chord connection to resist this relatively small perpendicular to grain force could have prevented this type of problem. Truss repair design was performed by the owners structural engineer.
Case Study 12. Albertsons Grocery Warehouse Portland, Oregon

This case study illustrates snow drifting. Albertsons Inc. maintains a grocery warehouse in Portland, Oregon, servicing retail stores in Oregon and Washington. The 14-acre warehouse built in 1987 is

15.54

CHAPTER FIFTEEN

framed with glulam timber beams and a panelized timber roof system. Attached to the primary building was a small 60-ft 80-ft recycling and salvage building. Walls of the primary building were 11 ft higher than the roof of the salvage building at the location where they joined. The 1994 UBC provides a design procedure for calculating the triangular surcharge of drifted snow load against an adjacent structure. Using that procedure, the calculated uniform snow load is 134.5 lb/ft2 adjacent to the building, tapering to the minimum snow load of 25 lb/ft2 in 25.5 ft, as shown in Fig. 15.19. Previous editions of the building code carried slightly different formulas yielding similar results. The roof of the adjacent building was apparently designed without snow drift consideration. During the winter of 19941995, wind-driven snow accumulated at the wall adjacent to the higher building, which acted as a snow fence. The full 11ft elevation filled with snow to the point where workers could walk from lower to upper roof level on the snow. This occurred in an area where snowfall is typically light but winds can be high. The excessive loading fractured one purlin and several subpurlins, overstressing the entire area, and they were shored. The forensic evaluation was conducted in July 1995, and a repair procedure was developed. Subsequently it was learned

yy ;; yy ;; yy ;; yy ;;
Wd = 25.5 FT 25 lb/ft2 60 FT

SNOW LOAD PER 1994 UBC

ACTUAL SNOWDRIFT

RECYCLING & SALVAGE BUILDING

yy ;; yy ;;

134.5 lb/ft2

Wb > 500 FT

11'-0"

WAREHOUSE BUILDING

BUILDING SECTION

FIGURE 15.19 Snow load surcharge against an adjacent structure per 1994 UBC. Albertsons Grocery Warehouse, Portland, Oregon.

TIMBER STRUCTURES

15.55

that Albertsons planned a major expansion which would require removal of the recycling and salvage building. As such, the shoring was retained and the building occupied until demolished by the facility expansion.
Case Study 13. Guilford of Maine Textile Mill Newport, Maine

This case study illustrates

G G

State-of-the-art criteria revisions Beams in bending

The textile mill structure was built in 1959 with a major addition in 1962. The building footprint is approximately 228 ft 176 ft. The roof structure is framed with 38 glulam beams 9 in 451 2 in in size. Allowable bending stress of the combination 11 glulam beams was 2600 lb/in2 per the 1957 NDS. Beam span is 85.5 ft with 12ft beam spacing. The original loading criteria were 20 lb/ft2 dead load and 40 lb/ft2 snow load. A design check per the original design criteria showed the following: Size factor CF Fb DL
SL

0.862 2557 lb/in2

7,895,000 in lb 3088 in3

where the snow load, SL, is 40 lb/ft2 Fb (2600 lb/in2)(1.15)(0.862) CD C F fb /Fb 2557 lb/in2 2577 lb/in2 2577 lb/in2

0.992

Flexural distress of one beam occurred in February 1998. The beam was shored and did not collapse. The authors firm was retained for forensic evaluation in March 1998. The allowable bending stress for combination 11 glulam was adjusted to 2050 lb/in2 based upon current allowable stress values for the grades of material present in the original beam layup. The dead load of 20 lb/ft2 was verified. The ground snow load increased to 100 lb/ft2 in this jurisdiction, adjusted to 70 lb/ft2 for roof snow load. A design check of the existing beam per present criteria showed the following: Volume factor CV fbDL
SL

0.719 3835 lb/in2

11,840,000 in lb 3088 in3

15.56

CHAPTER FIFTEEN

where the snow load, SL, is 70 lb/ft2 Fb (2050 lb/in2)(1.15)(0.719) CD CV fbDL SL Fb fbDL SL Fb
40

1695 lb/in2

2557 lb/in2 1695 lb/in2 3835 lb/in2 1695 lb/in2

1.51

70

2.26

Based upon 1998 design criteria, the beam capacity remaining for snow loading is 19.8 lb/ft2. The distressed beam will be shored permanently, and post tensioning or intermediate support was recommended for the remaining beams.
Case Study 14. Church of Christ Spokane, Washington

This case study illustrates

G G

Arch repair with moment connections The unexpected

The Church of Christ roof structure was framed with glulam Tudor arches of the type commonly used for small church sanctuaries. The arch bases measured 51 8 in 15 in. A deranged person with a vendetta against churches vandalized numerous churches in eastern Washington over time. The person gained entry to the church at night and attempted to destroy the building by fire. A propane torch was wired to the base of several arches and lighted. The bases of three arches were severely charred in the vicinity of the torches, but the full building did not burn. The remaining cross section of the most severely burned arch was approximately equivalent to a 4 4 member. For repair, the arch was shored against vertical load plus horizontal thrust. The lower 11 2 ft of arch leg was cut off and fitted with a steel shoe. The shoe acted as a moment connection in that compression was confined to a bearing plate at the inside edge of the arch, and tension was taken with a steel strap fastened to the back of the arch with shear plates and lag bolts. The steel replacement shoe was narrower than the 51 8-in-wide arch so that strips cut from a glulam beam could be fit on each side covering the steel and provided a cosmetic match with the glulam arch above.

Case Study 15. Hood River Valley High School Theater Hood River, Oregon

This case study illustrates

TIMBER STRUCTURES
G G

15.57

Secondary effects Cross-grain tension at web-chord truss connections

The theater structure was designed and built in 1977 with walls of tilt-up concrete and roof framing of light timber trusses. Truss length is 64 ft 111 2 in, truss depth is 45 in, and truss spacing is 32 in. The wood chord wood web trusses were of a unique design where web-to-chord and web-to-web connections consisted of glued finger joints. Roof collapse occurred on January 12, 1979, under snow loading well below the design snow load. Initial forensic investigation centered on the strength of adhesive bond at web-to-chord finger joints. The authors firm was retained for forensic evaluation by the school district. Several undamaged trusses were removed from the structure and tested to destruction on a test rack. Failure of the three trusses tested occurred well below design load with an average of dead load plus 56 percent of design snow load. The failure mode could be observed in the test rack. Failure occurred not in the adhesive but as cross-grain tension in the wood at web-to-chord finger joints. Nodal rotation of the truss connection caused a triangular-type tension distribution across the finger joint connection. The outer web connection failed, followed by inner web connections in rapid succession.

REFERENCES
1. National Design Specification for Wood Construction (NDS), American Forest & Paper Association, Washington, DC, ANSI-AF & PA NDS-1997 and previous editions. 2. ASTM D 3737-89a, Standard Method for Establishing Stresses for Structural Glued Laminated Timber, American Society for Testing and Materials, Philadelphia, PA. 3. AITC 500-91, Determination of Design Values for Structural Glued Laminated Timber in Accordance with ASTM D 3737-89a, American Institute of Timber Construction, Englewood, CO. 4. AITC 117-93 Design, Standard Specification for Structural Glued Laminated Timber of Softwood Species, American Institute of Timber Construction, Englewood, CO, AITC 117-93 with 1994 and 1998 addenda. 5. ASTM D 198-94, Standard Methods of Static Tests of Lumber in Structural Sizes, American Society for Testing and Materials, Philadelphia, PA. 6. Timber Construction Manual, American Institute of Timber Construction (AITC), Wiley, New York, 1994. 7. ANSI/AITC Standard A 190.1-1992 for Wood ProductsStructural Glued Laminated Timber, American Institute of Timber Construction, Englewood, CO. 8. Inspection Manual AITC 200-92 for Structural Glued Laminated Timber, American Institute of Timber Construction, Englewood, CO. 9. Wood Handbook: Wood as an Engineering Material, Reprinted from Forest Products Laboratory, General Technology Report FPL-GTR-113, Forest Products Society, Madison, WI.

15.58

CHAPTER FIFTEEN

10. ASTM D 143-83, Standard Methods of Testing Small Clear Specimens of Timbers, American Society for Testing and Materials, Philadelphia, PA. 11. ASTM D 4761-96, Standard Test Methods for Mechanical Properties of Lumber and Wood-Base Structural Material, American Society for Testing and Materials, Philadelphia, PA. 12. ASTM D 4442-92, Standard Test Methods for Direct Moisture Content Measurement of Wood and Wood-Base Material, American Society for Testing and Materials, Philadelphia, PA. 13. ASTM D 4444-92, Standard Test Methods for Use and Calibration of Hand-Held Moisture Meters, American Society for Testing and Materials, Philadelphia, PA. 14. Directive No. 29, National Emergency Specification for the Design, Fabrication and Erection of Stress Grade Lumber and Its Fastenings for Buildings, Conservation Division, War Production Board, Washington, August 9, 1943. 15. ASCE Standard 7-95, Minimum Design Loads for Buildings and Other Structures, American Society of Civil Engineers, Reston, VA, 1998. 16. Uniform Building Code, vol. 2, International Conference of Building Officials (ICBO), Whittier, CA, 1997. 17. Donald W. Neal, Restoration of Navy LTA (Lighter Than Air) Hangars, Evaluation and Upgrading of Wood Structures, Session of ASCE Structures Congress, New Orleans, LA, 1986.